[go: up one dir, main page]

WO2019102497A1 - Method and system for activity recording, visualisation and analysis for identified segments of forest - Google Patents

Method and system for activity recording, visualisation and analysis for identified segments of forest Download PDF

Info

Publication number
WO2019102497A1
WO2019102497A1 PCT/IN2018/050786 IN2018050786W WO2019102497A1 WO 2019102497 A1 WO2019102497 A1 WO 2019102497A1 IN 2018050786 W IN2018050786 W IN 2018050786W WO 2019102497 A1 WO2019102497 A1 WO 2019102497A1
Authority
WO
WIPO (PCT)
Prior art keywords
activity
edge node
sensor
forest
seismic
Prior art date
Application number
PCT/IN2018/050786
Other languages
French (fr)
Inventor
Ripul Ghosh
Aparna AKULA NAGA VARA
Siddhartha Sarkar
Anirudh VAJPEYI
Harish Kumar Sardana
Satish Kumar
Amarendra Goap
Original Assignee
Council Of Scientific And Industrial Research
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Council Of Scientific And Industrial Research filed Critical Council Of Scientific And Industrial Research
Publication of WO2019102497A1 publication Critical patent/WO2019102497A1/en

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
    • G01V1/001—Acoustic presence detection
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
    • G01V1/16—Receiving elements for seismic signals; Arrangements or adaptations of receiving elements
    • G01V1/18—Receiving elements, e.g. seismometer, geophone or torque detectors, for localised single point measurements
    • G01V1/189—Combinations of different types of receiving elements
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
    • G01V1/22—Transmitting seismic signals to recording or processing apparatus
    • G01V1/223—Radioseismic systems
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V2210/00—Details of seismic processing or analysis
    • G01V2210/10—Aspects of acoustic signal generation or detection
    • G01V2210/16—Survey configurations
    • G01V2210/169—Sparse arrays
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B13/00—Burglar, theft or intruder alarms
    • G08B13/16—Actuation by interference with mechanical vibrations in air or other fluid
    • G08B13/1654—Actuation by interference with mechanical vibrations in air or other fluid using passive vibration detection systems
    • G08B13/1663—Actuation by interference with mechanical vibrations in air or other fluid using passive vibration detection systems using seismic sensing means
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B27/00—Alarm systems in which the alarm condition is signalled from a central station to a plurality of substations

Definitions

  • Present invention relates to an apparatus and methodology based on edge computing platform for real-time monitoring of one or more specific ground activities such as movement of humans, vehicles, animals using primary seismic sensors in a defined geographical space, comprising of at least one sensor connected to an edge computing unit; capable of multi- sensor signal acquisition, digitizers, sleep management unit, activity analysis controller having intelligent algorithms for signal detection and classification of the activities; which is further connected to real-time decision recommender analytics server having artificial intelligence for trend analysis, actuation of other secondary modalities and multi-level alert generation to concerned group within a defined geographical forest area.
  • Automated activity monitoring is an emerging research area which deals with an application framework for automatic detection and classification of targets.
  • new algorithms which can mimic human brain like operations
  • people nowadays are developing systems where intelligence can be embedded for the purpose of decision making.
  • the real-world perception data is multi-dimensional which often takes time to decipher and visualize.
  • the decisions taken are subjective and can vary from one person to other. Often it is seen for the purpose of urbanisation, people have intruded inside the forest regions leading to expansion of nearby villages into forest regions or there are possibilities of linear infrastructural development such as roads, highways and railway lines cutting through forest regions. Thereby, disturbing the natural ecosystem of the animals and paving way towards conflict and threat situations for both animals and humans.
  • US patent 5452364 systems for conducting surveys of wildlife vocalization has been used. It comprises of a sound receiving unit which includes parabolic reflector having a focal point, three microphone elements, of which the center microphone is located at the reflector focal point, and the other two located at an equal offset diametrically, an audio processing circuit which analyses and process the signals to determine the direction of species sound.
  • the audio processing circuit consists of bandpass filter to selectively monitor the audio signals of interest.
  • the invention uses only microphones, which have a specific range and dependent on the vocalization and are more prone towards other environmental noises.
  • the invention uses only microphones as a sensor input to the system, which are dependent on vocalization of the species and are more prone towards other environmental noises.
  • US patent 8457962B2 showed a remote surveillance system with three microphones for recording audio data in the woods.
  • the device can be programmed to record at desired times of the day.
  • the device has playback capabilities for any recorded data and can be interfaced with personal computer via digital storage device.
  • a software for PC is provided which is capable of analyzing the data and provides the directionality and distance of sound source.
  • the invention is based on acoustics and uses of microphones shows a strong dependency on the vocalization of species.
  • an apparatus for automatic detection and monitoring of perimeter physical movement comprises of two fence posts which protects a subportion of the perimeter. Each of them is equipped with a motion sensor and a laser sensor.
  • the motion sensors automatically perform detection upon sensing any physical movement in the region outside the subportion.
  • the laser sensors are activated to track the physical activity.
  • the system is capable of detecting and tracking vehicles or humans.
  • the sensors used for primary detection have a limited field of view. Classifying various events has not been provided in the scope of this patent.
  • the sensors used in this invention are vision based and have a limited field of view.
  • the camera can have specific blind spots if the targets are being occluded by trees, plant canopy or other objects in the forest area.
  • the system has a strong dependency on the communication tower as the total processing is being done at the control server level.
  • Analytics server is missing from all the closest prior arts that can track the unknown noticeable seismic signature of an object based on artificial intelligence using a network of clusters of edge nodes.
  • Trend analytics server with decision recommendation is not present in existing systems for the effective monitoring/planning of human/animal activity in the forest area using seismic sensors.
  • the present invention focuses on providing an apparatus and methodology for generalised and secure ground activity monitoring framework in the identified segments of forests.
  • the objective of this invention is to provide a system and methodology for generalised and secure ground activity monitoring framework in the identified segments of forests.
  • Another objective of this invention is to provide a methodology and apparatus for detection of activities in a resource-constrained environment.
  • Another objective of this invention is to make the system capable of monitoring activities in spite of non-line-of-sight due to occlusions from trees or canopy in forest areas.
  • Another objective of this invention is to detect activity caused due to locomotion/movement irrespective in the absence of any vocalization/acoustic activity of the source.
  • Another objective of this invention is to maintain contiguous detection range by placing the seismic sensors uniformly or non-uniformly to suit a variety of terrains.
  • Another objective of this invention is to provide visualisation of user selectable short-term and long-term trend of the pre-defined and unknown activities.
  • Yet another objective of this invention is to present perception of the seismic footprints of detected activities with spatio-temporal information.
  • Another objective of this invention is to notify concerned authorities for the respective category of activities in real-time.
  • the present invention relates to a system and method based on edge computing platform for real-time monitoring, detection and resolving of a particular ground seismic activity caused due to movement of animals, humans and vehicles in a defined geographical space of a forest area.
  • the edge node is connected to at least one seismic senor and an analytics server for real-time alert generation and trend analytics.
  • the edge node comprises of an activity analysis controller with intelligent algorithms having a high probability of detection.
  • a plurality of edge nodes forms a cluster to monitor a contiguous range of forest area.
  • a plurality of clusters forms network of clusters to be used for sensing discontinuous regions in a forest area.
  • the edge nodes are further connected to an analytics server to serve trend analysis, activity information visualisation map and real-time decision support to disseminate warning messages to one or more specific concerned groups.
  • the trend analytic engine uses artificial intelligence for providing long-term and short-term trend of the activities.
  • FIG 1 illustrates an exemplary least possible configuration of the invention
  • FIG 2 illustrates a preferable edge node connected to sensor array
  • FIG 3 illustrates a preferable layout of the edge node with sensors showing hierarchical alert zones
  • FIG 4 illustrates a preferable layout of the linear placement of sensor with edge node deployed in an open area
  • FIG 5 illustrates a method for automatic activity detection and classification and visualisation through analytics server
  • FIG 6 illustrates an exemplary configuration roadway infrastructure passing though forest area
  • FIG 7 illustrates an exemplary configuration of protection of an agricultural crop field nearby forest area
  • FIG 8 illustrates a preferable visualisation of a cluster comprising of edge nodes showing spatio-temporal activities and alert mechanism
  • FIG 9 illustrates a preferable visualisation of hourly data trend of the activities
  • FIG 10 illustrates a preferable visualisation of historical trend hourly data trend of the activities
  • FIG 11 illustrates an exemplary GIS map showing network of clusters.
  • Method and system for activity recording, visualization and analysis for identified segments of forest comprises of seismic sensor 101, sensor cable 102, edge processing node 103 connected independently to alert dissemination devices 107.
  • edge processing node 103 In case of multiple nodes connected to monitor an extended zone, it will form a cluster 104 and a number of clusters of edge node further forms network of clusters 105.
  • These are connected to the analytics server 106 for visualisation, trend analytics and real-time alert/decisions dissemination via 107.
  • Fig. 1 illustrates the exemplary configuration of connected network and its elements.
  • the edge processing node 103 shown in Fig. 2 shows the architecture and processing embodiments.
  • Seismic sensor 101 or an array of seismic sensors 202 can be incorporated for multi-sensor signal acquisition 203, which comprises of dedicated analog filters 205 and simultaneously sampling analog-to-digital converters (ADC).
  • the sensors data can be further transferred to the edge node via wireless protocol.
  • Seismic signal is first processed through a low power wake-up-circuit 204 for a true detection in the neighbourhood of the sensor. On true detection 204 triggers the activity analysis controller 207 where the main application program 2015 does the data processing.
  • 2015 comprises of first multi-channel activity detection algorithm 209, second classification engine 2010 for classification of the detected activity and third the statistical analysis engine 2011 which computes the temporal trend of all the trigger details of the sensors.
  • the activity pattern analysis output gives feedback to the power, clock and sleep state management unit 201 to put 207 to sleep state if enough levels of activity is not detected. Highly synchronised time is maintained through global positioning system (GPS) and on-board real-time clock (RTC) 208. These are managed through 201.
  • GPS global positioning system
  • RTC real-time clock
  • 103 has a provision of directly disseminating alert information through communication and alert unit 2014 by means of GSM unit 2012 and audio-visual alert 2013. Additionally, 2014 also provides temporal trend information to analytics server 106 and it is further shared with other edge nodes 103 of 104 and 105.
  • the edge-processing node 103 is capable of interfacing more than one seismic sensor 101 and 202.
  • Each individual 101 and 202 has a circular sensing range 301 of radius d 302.
  • the sensing strength increases radially inward towards the sensor.
  • 101 and 202 are placed in such a way that they divide into hierarchical sensing zones to protect an area 306 along the forest boundary such as agricultural cropland, human settlements and transportation infrastructure.
  • Sensing zones are classified to as low alert zone 303, moderate alert zone 304 and high alert zone 305.
  • the data from the seismic sensor placed at 305 will not be processed and 103 will be in sleep state.
  • the sensors in region 304 and 305 gets activated afterwards. Consequently, the process reduces power consumption of the total system keeping few sensors active and others in sleep state.
  • Fig. 3 illustrates an exemplary configuration of sensor placement strategy for monitoring the protected zone area.
  • FIG. 4 Another sensor placement strategy is devised in Fig. 4 where an open area 402 of specific length is to be monitored to protect the zones 401 and 403 on both sides of 402. 101 and 202 are placed in a linear array along the 402 to sense and classify any activity happening in the location nearby.
  • Fig. 5 illustrates the method for real-time activity monitoring and analytics.
  • 501 denotes the user parameter settings block where the user can set their own parameters such as prior classification models 502, sampling rate of signal acquisition 503 and data recording mode 504.
  • the recording mode can be further configured into event based, continuous and scheduled.
  • the data from seismic sensors 101 connected to the edge processing node 103 is acquired through signal acquisition routine 5011.
  • 508 is the application routine which is embedded into 103.
  • Pre-processing and filtering of the raw signal is done through 507. Threshold computation is done adaptively at each iteration using 5012.
  • 506 confirms if the current signal level is greater than the current threshold value, and thus declaring a detected activity.
  • the current threshold is updated with a new threshold value for the next iteration.
  • a detection sustenance factor for a pre-defined time is considered in 505, this in turn provides feedback to the sleep controller 509.
  • the classification routine 5025 is initiated.
  • the triggering details are locally stored in 5023 and classified activity is stored in 5024.
  • the overall power management unit 5010 manages the resources and optimizes power usage. Time synchronisation is done at 208 using RTC and GPS. Time-stamped information from 5023 and 5024 are communicated to the analytics server 106, where information from other edge nodes 103 and clusters 104 are available.
  • the statistical analysis and visualisation is being done by various routines 5013, 5014, 5017.
  • Alert dissemination unit 5018 comprises of audio/visual alert/decision dissemination 5019, SMS 5020, email alert 5021; and 5022 actuates secondary sensing modalities. 508 also sends information regarding strongly triggered activities which could not be associated with pre-defined classes with required confidence levels, so as to enable analytic server 106 run autonomous/manual artificial intelligence routines 5015 to update models 5016 with new data and/or activity categories.
  • Fig. 6 shows an exemplary configuration where a transportation infrastructure is passing through a forest area 601.
  • the example shows here that the movement of animals 605 can be tracked using a single seismic sensor 101 or an extended range sensing using seismic array 202.
  • the seismic sensor array 202 is buried under the soil to conceal and protect it from animals and from trespassers. The inherent layout of sensors ensures safety from theft.
  • the forest area is virtually divided into three zones namely low alert 303, moderate alert 304 and high alert 305.
  • the conflict region 306 in this example is a highway infrastructure where moving vehicles 603 is allowed to pass the forest zone with an early alert message if there is any animal presence 605, thus reducing human-animal conflict scenarios.
  • the seismic array is further connected to the edge node 103 which is solar powered by 602.
  • the signal received is processed at the edge node 103 and the key information is transferred to the analytics server 106 via 604.
  • the visualisation 5017 at the analytics server 106 shows a spatio-temporal activity map of a cluster showing classified activities in Fig. 8.
  • the alert mechanism unit 5018 is further used to send alert via audio/visual 5019, email 5020, SMS 5021 and activating other secondary sensors using contact relays 5022 upon assessing the vulnerability of the detected activities.
  • the visualizer 5017 shows a cumulative detected activity trend chart 902, showing hourly data of each day 903 with pop-up window of current activity, if any. Each hour data is shown via 901.
  • the analytics further shows the current trend of the detected activities at each hour of a day via 5014.
  • Fig. 10 further shows of retrieving historical trend through 1001.
  • Each hour data further shows the count of detected activity in each of predefined class 5013 and unknown activity 5015.
  • Fig. 11 shows the exemplary network of cluster 105 having individual cluster 104 in identified segment forest. Further, spatio-temporal detection visualisation map with classified detected activities are shown in Fig. 11.
  • Fig. 7 shows an exemplary configuration where an agricultural crop field in a village 702 located along the area adjoining forest 601.
  • the example shows a device generating early alerts via 5018, while animals are approaching towards the crop filed or village areas located in vicinity of forest areas.
  • the movement of animals 605 inside forest is sensed using seismic sensor 101 or seismic array 202.
  • the seismic sensor array 202 is buried under the soil for protecting from animal movement.
  • Two seismic arrays 202 are shown in this example for hierarchical sensing of any activity happening inside forest area. Any animal movement occurring far away from the crop land is designated as a low alert area 303. Movement of animals happening near the crop land triggers the edge nodes 103 connected to the sensor arrays 202.
  • the intelligent algorithms 508 detects and classifies the detected signal into predefined classes 603, 605 and 701.
  • the key information is communicated to the analytics server 106 for real-time decision support.
  • the alert message is communicated to the farmer or any village residence 701 via 5018 shown in Fig. 8.
  • the edge node is kept near the village/ residential area of the farmers and are all solar powered by 602.
  • the system and method has capabilities of edge processing of different ground activity having higher detection probabilities.
  • the primary seismic sensor array is buried under the soil to protect it from animals and trespassers.
  • the inherent layout of sensors ensures safety from theft and tampering.
  • Seismic sensor ensures omnidirectional sensing with no blind zone in case of contiguous terrain of forest.
  • Seismic sensors also ensure detection of activity caused due to locomotion/movement even in the absence of any vocalization/acoustic activity of source.
  • pluraliturality of spatially distributed sensors can increase probability and range of detection using edge computing on node level.
  • the sensor locations and spacing is fixed uniformly or non-uniformly to adapt variability of terrains and maintain contiguous detection range.
  • the locus of all the sensors is chosen as per the desired profile of the zones and a priori knowledge of activities in the specified zones.
  • the edge node comprises of activity analysis controller with intelligent edge processing algorithms and further power and sleep management unit for optimization of power resource.
  • pluraliturality of edge nodes, clusters and network of clusters are further connected to analytics server for real-time decision support and trend analysis.
  • the alert generation has been configured in zone wise to issue hierarchal multi level warning mechanism.
  • Activity detected are locally stored and processed in the edge level node and the triggered information is passed on to the analytics server.
  • the analytics server provides trend analysis of activities and decision recommendation for the effective planning of human activity in the forest.
  • the analytics server can also track the unknown noticeable seismic signature of an object based on artificial intelligence.

Landscapes

  • Physics & Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Environmental & Geological Engineering (AREA)
  • Geology (AREA)
  • Remote Sensing (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Geophysics (AREA)
  • Alarm Systems (AREA)

Abstract

The present invention relates to an intelligent framework for seismic activity monitoring in a defined geographical area of a forest that is capable of visualisation, analysis, multi-level actuations and information broadcasting to a number of multi-type user bases, wherein, intelligent framework for real-time seismic activity monitoring is accomplished using at least one edge computing platform and an analytics server enabled with artificial intelligence; wherein, the real-time seismic activity monitoring comprises of at least one of the methods, detection; tracking; localisation; categorisation of activities caused due to movement of one or more target seismic sources such as animals, humans and vehicles; wherein information to be broadcasted can be data, statistics, inference, warning, etc.; wherein information broadcasting is through at least one of the dissemination methods, audio-visual alert; short message service; electronic mail; push notification; wherein the users belong to at least one of the user groups, forest officials; security forces; traffic management personnel; railway operators or officials; people residing in the nearby vicinity; commuters in that region.

Description

METHOD AND SYSTEM FOR ACTIVITY RECORDING, VISUALISATION AND ANALYSIS FOR IDENTIFIED SEGMENTS OF FOREST
FIELD OF THE INVENTION
Present invention relates to an apparatus and methodology based on edge computing platform for real-time monitoring of one or more specific ground activities such as movement of humans, vehicles, animals using primary seismic sensors in a defined geographical space, comprising of at least one sensor connected to an edge computing unit; capable of multi- sensor signal acquisition, digitizers, sleep management unit, activity analysis controller having intelligent algorithms for signal detection and classification of the activities; which is further connected to real-time decision recommender analytics server having artificial intelligence for trend analysis, actuation of other secondary modalities and multi-level alert generation to concerned group within a defined geographical forest area.
BACKGROUND OF THE INVENTION
Automated activity monitoring is an emerging research area which deals with an application framework for automatic detection and classification of targets. With the development of new algorithms, which can mimic human brain like operations, people nowadays are developing systems where intelligence can be embedded for the purpose of decision making. The real-world perception data is multi-dimensional which often takes time to decipher and visualize. The decisions taken are subjective and can vary from one person to other. Often it is seen for the purpose of urbanisation, people have intruded inside the forest regions leading to expansion of nearby villages into forest regions or there are possibilities of linear infrastructural development such as roads, highways and railway lines cutting through forest regions. Thereby, disturbing the natural ecosystem of the animals and paving way towards conflict and threat situations for both animals and humans. Therefore, it is inevitably necessary to monitor activities all through the day and night inside these regions which are complex and a laborious task. Human patrolling is not feasible or very difficult at times. Therefore, to overcome the above-mentioned challenges, a system for automated monitoring, which involves detection, classification, visualisation and analysis has been sought in this invention. Augmenting these measures with a technological solution to play the role of sensing activities such as person and vehicle movement in an identified area of a forest region, animal movement towards urban area or linear infrastructures and subsequently alerting the concerned persons would be of great advantage.
In US patent 5452364, systems for conducting surveys of wildlife vocalization has been used. It comprises of a sound receiving unit which includes parabolic reflector having a focal point, three microphone elements, of which the center microphone is located at the reflector focal point, and the other two located at an equal offset diametrically, an audio processing circuit which analyses and process the signals to determine the direction of species sound. The audio processing circuit consists of bandpass filter to selectively monitor the audio signals of interest.
The invention uses only microphones, which have a specific range and dependent on the vocalization and are more prone towards other environmental noises.
In US patent 5956463, an automated system for monitoring wildlife vocalization and recording the same for further analysing and identification is shown. The system minimizes the need for human intervention to save labour and cost of operation. The recorded data are pre-processed, windowed to extract features which are used as an input to a neural network for class identification.
The invention uses only microphones as a sensor input to the system, which are dependent on vocalization of the species and are more prone towards other environmental noises.
US patent 8457962B2 showed a remote surveillance system with three microphones for recording audio data in the woods. The device can be programmed to record at desired times of the day. The device has playback capabilities for any recorded data and can be interfaced with personal computer via digital storage device. A software for PC is provided which is capable of analyzing the data and provides the directionality and distance of sound source.
The invention is based on acoustics and uses of microphones shows a strong dependency on the vocalization of species.
In US patent 6816073B2, an apparatus for automatic detection and monitoring of perimeter physical movement is shown. It comprises of two fence posts which protects a subportion of the perimeter. Each of them is equipped with a motion sensor and a laser sensor. The motion sensors automatically perform detection upon sensing any physical movement in the region outside the subportion. Upon detecting the physical movement in the region by the motion sensors, the laser sensors are activated to track the physical activity.
The system is capable of detecting and tracking vehicles or humans. The sensors used for primary detection have a limited field of view. Classifying various events has not been provided in the scope of this patent.
In Indian patent IN2010MU02170, a real-time anti-poaching surveillance and wildlife tracking system using long-distance sensing towers and short distance sensing towers is used. The sensing towers are connected to a web cloud via hopping towers to the control center. The total processing of the data is being done at control center. The system shows the capability of sensing activities related to animal poaching.
The sensors used in this invention are vision based and have a limited field of view. The camera can have specific blind spots if the targets are being occluded by trees, plant canopy or other objects in the forest area. The system has a strong dependency on the communication tower as the total processing is being done at the control server level.
From the closest prior art, following gaps and drawbacks are evident:
• Existing systems for this purpose did not use seismic sensors for ground activity monitoring in forest areas.
• The aforesaid systems also do not process the signals at edge level. The signals are completely transferred to the control server for further processing.
• The sensors in the existing systems are exposed in nature due to their inherent mode of operation.
• The forest being naturally obstructing in nature, creates hindrance in performance for line-of- sight sensors.
• Analytics server is missing from all the closest prior arts that can track the unknown noticeable seismic signature of an object based on artificial intelligence using a network of clusters of edge nodes.
• Trend analytics server with decision recommendation is not present in existing systems for the effective monitoring/planning of human/animal activity in the forest area using seismic sensors. The present invention focuses on providing an apparatus and methodology for generalised and secure ground activity monitoring framework in the identified segments of forests.
OBJECTIVES OF THE INVENTION
The objective of this invention is to provide a system and methodology for generalised and secure ground activity monitoring framework in the identified segments of forests.
Another objective of this invention is to provide a methodology and apparatus for detection of activities in a resource-constrained environment.
Another objective of this invention is to make the system capable of monitoring activities in spite of non-line-of-sight due to occlusions from trees or canopy in forest areas.
Another objective of this invention is to detect activity caused due to locomotion/movement irrespective in the absence of any vocalization/acoustic activity of the source.
Another objective of this invention is to maintain contiguous detection range by placing the seismic sensors uniformly or non-uniformly to suit a variety of terrains.
Another objective of this invention is to provide the system artificial intelligence capable of categorising activities into pre-defined types. Yet another objective of this invention is to provide the system artificial intelligence capable of automatically associating unknown repeatedly occurring patterns into new activity types.
Another objective of this invention is to provide visualisation of user selectable short-term and long-term trend of the pre-defined and unknown activities.
Yet another objective of this invention is to present perception of the seismic footprints of detected activities with spatio-temporal information.
Another objective of this invention is to notify concerned authorities for the respective category of activities in real-time.
SUMMARY OF THE INVENTION
The present invention relates to a system and method based on edge computing platform for real-time monitoring, detection and resolving of a particular ground seismic activity caused due to movement of animals, humans and vehicles in a defined geographical space of a forest area. The edge node is connected to at least one seismic senor and an analytics server for real-time alert generation and trend analytics. The edge node comprises of an activity analysis controller with intelligent algorithms having a high probability of detection. A plurality of edge nodes forms a cluster to monitor a contiguous range of forest area. A plurality of clusters forms network of clusters to be used for sensing discontinuous regions in a forest area. The edge nodes are further connected to an analytics server to serve trend analysis, activity information visualisation map and real-time decision support to disseminate warning messages to one or more specific concerned groups. The trend analytic engine uses artificial intelligence for providing long-term and short-term trend of the activities.
BRIEF DESCRIPTION OF THE DRAWING
FIG 1 illustrates an exemplary least possible configuration of the invention;
FIG 2 illustrates a preferable edge node connected to sensor array;
FIG 3 illustrates a preferable layout of the edge node with sensors showing hierarchical alert zones;
FIG 4 illustrates a preferable layout of the linear placement of sensor with edge node deployed in an open area;
FIG 5 illustrates a method for automatic activity detection and classification and visualisation through analytics server;
FIG 6 illustrates an exemplary configuration roadway infrastructure passing though forest area;
FIG 7 illustrates an exemplary configuration of protection of an agricultural crop field nearby forest area;
FIG 8 illustrates a preferable visualisation of a cluster comprising of edge nodes showing spatio-temporal activities and alert mechanism;
FIG 9 illustrates a preferable visualisation of hourly data trend of the activities;
FIG 10 illustrates a preferable visualisation of historical trend hourly data trend of the activities;
FIG 11 illustrates an exemplary GIS map showing network of clusters. DETAILED DESCRIPTION OF THE INVENTION
Method and system for activity recording, visualization and analysis for identified segments of forest comprises of seismic sensor 101, sensor cable 102, edge processing node 103 connected independently to alert dissemination devices 107. In case of multiple nodes connected to monitor an extended zone, it will form a cluster 104 and a number of clusters of edge node further forms network of clusters 105. These are connected to the analytics server 106 for visualisation, trend analytics and real-time alert/decisions dissemination via 107. Fig. 1 illustrates the exemplary configuration of connected network and its elements. The edge processing node 103 shown in Fig. 2 shows the architecture and processing embodiments. Seismic sensor 101 or an array of seismic sensors 202 can be incorporated for multi-sensor signal acquisition 203, which comprises of dedicated analog filters 205 and simultaneously sampling analog-to-digital converters (ADC). The sensors data can be further transferred to the edge node via wireless protocol. Seismic signal is first processed through a low power wake-up-circuit 204 for a true detection in the neighbourhood of the sensor. On true detection 204 triggers the activity analysis controller 207 where the main application program 2015 does the data processing. 2015 comprises of first multi-channel activity detection algorithm 209, second classification engine 2010 for classification of the detected activity and third the statistical analysis engine 2011 which computes the temporal trend of all the trigger details of the sensors. The activity pattern analysis output gives feedback to the power, clock and sleep state management unit 201 to put 207 to sleep state if enough levels of activity is not detected. Highly synchronised time is maintained through global positioning system (GPS) and on-board real-time clock (RTC) 208. These are managed through 201. 103 has a provision of directly disseminating alert information through communication and alert unit 2014 by means of GSM unit 2012 and audio-visual alert 2013. Additionally, 2014 also provides temporal trend information to analytics server 106 and it is further shared with other edge nodes 103 of 104 and 105.
The edge-processing node 103 is capable of interfacing more than one seismic sensor 101 and 202. Each individual 101 and 202 has a circular sensing range 301 of radius d 302. The sensing strength increases radially inward towards the sensor. 101 and 202 are placed in such a way that they divide into hierarchical sensing zones to protect an area 306 along the forest boundary such as agricultural cropland, human settlements and transportation infrastructure. Sensing zones are classified to as low alert zone 303, moderate alert zone 304 and high alert zone 305. Initially, the data from the seismic sensor placed at 305 will not be processed and 103 will be in sleep state. As the activity is detected in 303 region the sensors in region 304 and 305 gets activated afterwards. Consequently, the process reduces power consumption of the total system keeping few sensors active and others in sleep state. Fig. 3 illustrates an exemplary configuration of sensor placement strategy for monitoring the protected zone area.
Another sensor placement strategy is devised in Fig. 4 where an open area 402 of specific length is to be monitored to protect the zones 401 and 403 on both sides of 402. 101 and 202 are placed in a linear array along the 402 to sense and classify any activity happening in the location nearby.
Fig. 5 illustrates the method for real-time activity monitoring and analytics. 501 denotes the user parameter settings block where the user can set their own parameters such as prior classification models 502, sampling rate of signal acquisition 503 and data recording mode 504. The recording mode can be further configured into event based, continuous and scheduled. The data from seismic sensors 101 connected to the edge processing node 103 is acquired through signal acquisition routine 5011. 508 is the application routine which is embedded into 103. Pre-processing and filtering of the raw signal is done through 507. Threshold computation is done adaptively at each iteration using 5012. 506 confirms if the current signal level is greater than the current threshold value, and thus declaring a detected activity. If activity is not detected 5012, then the current threshold is updated with a new threshold value for the next iteration. To reduce the false alarm, a detection sustenance factor for a pre-defined time is considered in 505, this in turn provides feedback to the sleep controller 509. For a true output of 505, the classification routine 5025 is initiated. The triggering details are locally stored in 5023 and classified activity is stored in 5024. The overall power management unit 5010 manages the resources and optimizes power usage. Time synchronisation is done at 208 using RTC and GPS. Time-stamped information from 5023 and 5024 are communicated to the analytics server 106, where information from other edge nodes 103 and clusters 104 are available. The statistical analysis and visualisation is being done by various routines 5013, 5014, 5017. Alert dissemination unit 5018 comprises of audio/visual alert/decision dissemination 5019, SMS 5020, email alert 5021; and 5022 actuates secondary sensing modalities. 508 also sends information regarding strongly triggered activities which could not be associated with pre-defined classes with required confidence levels, so as to enable analytic server 106 run autonomous/manual artificial intelligence routines 5015 to update models 5016 with new data and/or activity categories.
Working Example 1
Fig. 6 shows an exemplary configuration where a transportation infrastructure is passing through a forest area 601. The example shows here that the movement of animals 605 can be tracked using a single seismic sensor 101 or an extended range sensing using seismic array 202. The seismic sensor array 202 is buried under the soil to conceal and protect it from animals and from trespassers. The inherent layout of sensors ensures safety from theft. The forest area is virtually divided into three zones namely low alert 303, moderate alert 304 and high alert 305. The conflict region 306 in this example is a highway infrastructure where moving vehicles 603 is allowed to pass the forest zone with an early alert message if there is any animal presence 605, thus reducing human-animal conflict scenarios. The seismic array is further connected to the edge node 103 which is solar powered by 602. Upon any ground activity in the vicinity of the seismic array, the signal received is processed at the edge node 103 and the key information is transferred to the analytics server 106 via 604. The visualisation 5017 at the analytics server 106 shows a spatio-temporal activity map of a cluster showing classified activities in Fig. 8. The alert mechanism unit 5018 is further used to send alert via audio/visual 5019, email 5020, SMS 5021 and activating other secondary sensors using contact relays 5022 upon assessing the vulnerability of the detected activities. Further in Fig. 9 the visualizer 5017 shows a cumulative detected activity trend chart 902, showing hourly data of each day 903 with pop-up window of current activity, if any. Each hour data is shown via 901. The analytics further shows the current trend of the detected activities at each hour of a day via 5014. Fig. 10 further shows of retrieving historical trend through 1001. Each hour data further shows the count of detected activity in each of predefined class 5013 and unknown activity 5015. Fig. 11 shows the exemplary network of cluster 105 having individual cluster 104 in identified segment forest. Further, spatio-temporal detection visualisation map with classified detected activities are shown in Fig. 11. Working Example 2
Fig. 7 shows an exemplary configuration where an agricultural crop field in a village 702 located along the area adjoining forest 601. The example shows a device generating early alerts via 5018, while animals are approaching towards the crop filed or village areas located in vicinity of forest areas. The movement of animals 605 inside forest is sensed using seismic sensor 101 or seismic array 202. The seismic sensor array 202 is buried under the soil for protecting from animal movement. Two seismic arrays 202 are shown in this example for hierarchical sensing of any activity happening inside forest area. Any animal movement occurring far away from the crop land is designated as a low alert area 303. Movement of animals happening near the crop land triggers the edge nodes 103 connected to the sensor arrays 202. The intelligent algorithms 508 detects and classifies the detected signal into predefined classes 603, 605 and 701. The key information is communicated to the analytics server 106 for real-time decision support. The alert message is communicated to the farmer or any village residence 701 via 5018 shown in Fig. 8. The edge node is kept near the village/ residential area of the farmers and are all solar powered by 602.
ADVANTAGES
The main advantages of this apparatus are:
1. The system and method has capabilities of edge processing of different ground activity having higher detection probabilities.
2. The primary seismic sensor array is buried under the soil to protect it from animals and trespassers. The inherent layout of sensors ensures safety from theft and tampering.
3. Seismic sensor ensures omnidirectional sensing with no blind zone in case of contiguous terrain of forest.
4. Seismic sensors also ensure detection of activity caused due to locomotion/movement even in the absence of any vocalization/acoustic activity of source.
5. Plurality of spatially distributed sensors can increase probability and range of detection using edge computing on node level. 6. The sensor locations and spacing is fixed uniformly or non-uniformly to adapt variability of terrains and maintain contiguous detection range.
7. The locus of all the sensors is chosen as per the desired profile of the zones and a priori knowledge of activities in the specified zones.
8. The edge node comprises of activity analysis controller with intelligent edge processing algorithms and further power and sleep management unit for optimization of power resource.
9. Single edge node and neighbourhood signal processing achieves reliable performance through decision feedback between neighbouring nodes or within cluster of nodes, even in situations where a sensor is non-functional.
10. Plurality of edge nodes, clusters and network of clusters are further connected to analytics server for real-time decision support and trend analysis.
11. The alert generation has been configured in zone wise to issue hierarchal multi level warning mechanism.
12. Activity detected are locally stored and processed in the edge level node and the triggered information is passed on to the analytics server.
13. Multiple nodes and cluster of nodes ensures increased range of sensing geographically and reliability in tracking activities in real-time.
14. Early alert generation via analytics server to specific concerned group.
15. The analytics server provides trend analysis of activities and decision recommendation for the effective planning of human activity in the forest.
16. The analytics server can also track the unknown noticeable seismic signature of an object based on artificial intelligence.

Claims

We claim:
1. A method to provide intelligent framework for secure ground activity monitoring in identified segments of forest, comprising:
sensing by at least one sensor (101), at-least a seismic -pattern, wherein the sensor, which is connected to the edge node, is triggered due to any ground activity at its vicinity (302);
utilizing, by an edge node (103), artificial intelligence routines (2010) capable of categorising activities into pre-defined types, primarily movement of human (701), vehicles (603) and animals (605); wherein the edge node uses (103) recent activity pattern analysis feedback to power, clock and sleep state management unit (201) to put an activity analysis controller to sleep state (509) to optimize power resources;
storing and analysing outputted data from the edge node (103) by an analytics server (106) for visualizing, localising, tracking the acivities and generating alert thereupon.
2. The method in claim 1, wherein the artificial intelligence routines refer multiple trained models (502) of activities for categorisation (5013).
3. The method in claims 1-2, wherein the activity models are trained upon signature from at least one of the primary or secondary sensors.
4. The method in claim 1-3, wherein the analytic- server is configured to perform at least one of:
employ artificial intelligence routines to automatically associate unknown repeatedly occurring patterns (5015) into a new activity types (5016);
using artificial intelligence and trend analytics engine (5014) to visualize (5017) the seismic pattern detected at each edge node;
correlating the new activity types associated over time by secondary sensor modalities for activity recognition;
adding further the new activities to the list of pre-defined categories (5013) of selected edge nodes; using decision feedback mechanism between neighboring nodes or within cluster of nodes, even in situations where a sensor is non-functional;
using real-time decision support to categorize the detected activity into multi-class, multi-level, and disseminating the warning to a specific concerned group;
adapting the warning level from long-term trend and short-term trend of the activity using historical data (1001);
recommending probable spatio-temporal decision alert and actions from a pool of recommendations.
5. An apparatus to provide intelligent framework for secure ground activity monitoring in identified segments of forest, comprising at least one
a seismic sensor (101), wherein the sensor (101) is connected to the edge node and triggered due to any ground activity at its vicinity (302);
an edge-node (103) for signal processing of data from at least one sensor (101) at edge level capable of categorising activities into pre-defined types, primarily movement of human (701), vehicles (603) and animals (605); wherein the edge node uses (103) recent activity pattern analysis feedback to power, clock and sleep state management unit (201) to put an activity analysis controller (207) to sleep state (509) to optimize power resources ; and
an analytics server (106) operatively coupled to the edge node for information storage, analysis, visualisation and alert (107) generation, wherein the server (106) facilitates visualisation of the activity information (5017) based on output of the edge node and thereby provide a real-time decision support to disseminate warning message (5018) to one or more group.
6. The apparatus of claim 5, wherein the edge node (103) interprets ground activity data from at least one of the six types of primary seismic sensor (101) buried beneath soil: (I) vertical geophones, (II) borehole geophones, (III) accelerometer, (IV) seismometer, (V) hydrophone, (VI) geophone string.
7. The apparatus of claims 5-6, wherein the primary sensors can be further augmented by secondary sensors, optical; acoustic; electromagnetic and mechanical sensors for analytics establishment of each new seismic pattern observed.
8. The apparatus of claims 5-7, wherein overlapped placement aware signal processing at edge level provides false alarm rate reduction and enhancement of confidence of classification via redundant sensing (303, 304, 305).
9. The apparatus in claims 5-8, wherein the analytics server (106) uses artificial intelligence and trend analytics engine (5014) on the data received from each edge node (103), a cluster (104) and network of clusters (105) in a particular geographical area to serve activity information visualisation (5017) and real-time decision support to disseminate warning message (5018) to specific concerned group.
10. The apparatus in claims 5-9, wherein the trend analytics engine (5014) uses artificial intelligence and data interpretation method to visualize (902) the detected/classified activity via each edge node (5013), cluster and network of clusters in a particular geographical area providing long-term and short-term trend (5014). wherein the visualisation uses geographic information system (GIS) map (5017, Fig. 8) to show the overlaid sensors of the edge node(s) where the particular activity is being detected and classified;
wherein the dissemination engine uses at least any of the warning methods (5018), audio visual alert (5019); electronic mail (5020); short message service (5021); push notification; and
wherein edge node (103) and analytics server (106) are used for behavioural-monitoring of different species at different spatio-temporal scale.
wherein long-term behavioural monitoring of different species at analytics server is used to predict spatio-temporal likelihood of human-wildlife conflict and activities within monitoring zones.
PCT/IN2018/050786 2017-11-27 2018-11-26 Method and system for activity recording, visualisation and analysis for identified segments of forest WO2019102497A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
IN201711042427 2017-11-27
IN201711042427 2017-11-27

Publications (1)

Publication Number Publication Date
WO2019102497A1 true WO2019102497A1 (en) 2019-05-31

Family

ID=66630567

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/IN2018/050786 WO2019102497A1 (en) 2017-11-27 2018-11-26 Method and system for activity recording, visualisation and analysis for identified segments of forest

Country Status (1)

Country Link
WO (1) WO2019102497A1 (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090237241A1 (en) * 2008-03-19 2009-09-24 Wildlife Acoustics, Inc. Apparatus for scheduled low power autonomous data recording
WO2014129966A1 (en) * 2013-02-25 2014-08-28 Biomachines Pte Ltd Data logger and sentinel platform for sensor network

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090237241A1 (en) * 2008-03-19 2009-09-24 Wildlife Acoustics, Inc. Apparatus for scheduled low power autonomous data recording
WO2014129966A1 (en) * 2013-02-25 2014-08-28 Biomachines Pte Ltd Data logger and sentinel platform for sensor network

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
PRABU, M.: "An efficient surveillance system to detect elephant intrusion into forest borders using seismic sensors", INT J ADV ENGG TECH, vol. VII, no. I, January 2016 (2016-01-01), pages 1 - 6, XP055618476 *
WOOD, JASON D. ET AL.: "Methodological insights: Using seismic sensors to detect elephants and other large mammals: A potential census technique", JOURNAL OF APPLIED ECOLOGY, vol. 42, no. 3, 9 June 2005 (2005-06-09), pages 587 - 594, XP055618463 *

Similar Documents

Publication Publication Date Title
US11867539B2 (en) Acoustic method and system for providing digital data
US20240394573A1 (en) Joint processing for embedded data inference
EP2814012B1 (en) Cooperative intrusion detection
Allen et al. The MyShake platform: a global vision for earthquake early warning
CA3043400C (en) Systems and methods for detecting flying animals
KR102090442B1 (en) Ecological Environment Research System Based on Big Data
US9461846B2 (en) Multilayered configurable data fusion systems and methods for power and bandwidth efficient sensor networks
US10163315B2 (en) Localized hazard alert system and method of use
KR102531272B1 (en) Control device using artificial intelligence learning images and vibrations and intrusion alert systme including the same
US7656288B2 (en) Systems and methods for automatic proactive pattern recognition at a control center database
Sugumar et al. An improved real time image detection system for elephant intrusion along the forest border areas
CN108604403A (en) Event prediction system
US7710265B2 (en) Systems and methods for dynamic situational signal processing for target detection and classification
US10062255B1 (en) VMD fused radar—a hyper-volumetric ultra-low NAR sensor system
Dorfling et al. Satellite, aerial, and ground sensor fusion experiment for management of elephants and rhinos and poaching prevention
Nakandala et al. Detecting wild elephants via WSN for early warning system
CN120236360A (en) Area intrusion judgment method, system, device and medium for outdoor complex scene
AU2012333064B2 (en) Intrusion detection system
WO2019102497A1 (en) Method and system for activity recording, visualisation and analysis for identified segments of forest
US20230124662A1 (en) System and method for capturing an event of random occurance and length from a stream of continuous input data
Siewert et al. Acoustic, Seismic, and Visual Camera Sensor Fusion Experiments for Large Animal Detection and Tracking with Scalability
Aruchamy et al. Multi-zone fence perimeter surveillance: A new edge-fog architecture for efficient detection and classification of intrusion
Lynam et al. The rising tide of conservation technology: empowering the fight against poaching and unsustainable wildlife harvest
KR101857131B1 (en) Method for tracing location of surveillance assets
Rowe et al. Networks of simple sensors for detecting emplacement of improvised explosive devices

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18882141

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18882141

Country of ref document: EP

Kind code of ref document: A1