ElephantEdge aims to protect species from poaching via advanced tracking collar

A developer community called Hackster.io and Smart Parks are heading up an unusual technology collaboration with conservation groups to build an advanced tracking collar to protect elephants in the wild.

A design competition called ElephantEdge Challenge asks developers to pair software and machine learning (ML) models with new collar hardware containing a panoply of integrated circuits that are low power and provide wireless connectivity over Bluetooth 5, Thread and Zigbee. The collars are packed with sensors for audio pickup, location and position as well as low-power, wide-area antennas.

Conservationists believe extinction of the species, now estimated at 415,000 elephants in Africa, is possible within a decade due primarily to poaching.  The amount of poaching is on the rise because COVID-19 has reduced funding for law enforcement in wildlife areas, according to reports from conservationists.   In June, six elephants were killed in one day in Ethiopia’s Mago National Park, which compares to 10 in that nation for all of 2019.

The challenge is to build ML models with Edge Impulse Studio and a tracking dashboard with Avnet’s IoTConnect.  The software will be deployed on 10 production-grade collars manufactured by Institute IRNAS and deployed in various locations by Smart Parks, a Dutch enterprise that already relies on sending data about wild animal movements via LoRaWAN. Hackster.io is an Avnet developer community.

Tech collaborators on the project include Microsoft, Western Digital, Nordic Semiconductors, Taoglas, u-blox, and Vulcan EarthRanger, along with semiconductors from STMicroelectronics.

With open source software and the Azure IoT partner ecosystem, ElephantEdge is an example of collaboration “using the best technology has to offer to conserve and reverse the devastating loss of elephants,” said Sarah Maston, solution architect at Microsoft, in a statement.

 The top five tracking dashboards in the competition and the top five machine learning models will receive $500 in prizes.

The ML models will create a Human to Elephant Language, running on TinyML, that is designed to send real-time alerts to park rangers of when an elephant is moving into a high risk area.  Or, an alert might say when an elephant is heading to an area where farmers live, detecting for human presence by scanning for mobile phones or Wi-Fi hotspots nearby.  Another monitoring mode would look for aggressive elephant behavior by relying on motion and acoustic sensors on the collar.   Also, an onboard microphone could be used to listen for vocal communications between elephants.   

The challenge documentation notes that elephants are afraid of bees, speak with their feet and have poor hearing.The OpenCollar initiative for wildlife monitoring was first launched in early 2019. 

RELATED: Wearable gas sensor monitors health, environment