Portfolio sections
Portfolio Alexandre Kiss

↑↓ Navigate · Enter Open · Esc Close

product · Sept – Dec 2024

ECO-SENTINEL: an acoustic wildlife monitor

An acoustic monitor to help count the endangered capercaillie, developed with the Swiss Ornithological Institute and rangers of the Haute Chaîne du Jura.

Illustration of a calling capercaillie surrounded by tree-mounted monitoring boxes, with radio detections displayed on a map.
Each monitoring box classifies sounds locally and sends detected calls over LoRa. The dashboard records which box heard a call and when. The forest, box spacing and sound propagation are simplified for illustration.

Summary view

Project details

  1. walking survey bird falls silent

    Monitoring the endangered capercaillie

    The capercaillie, or grand tétras, is an endangered forest bird. The project presentation reports a 75 % decline in France. In the Haute Chaîne du Jura, reserve rangers and the Swiss Ornithological Institute monitor the population. During the survey season, rangers camp in the mountains two days a week to listen for calls.

  2. capercaillie rangers, camping boxes on the trees same call, three places one bird the singing month 2 days a week every day schematic

    Estimating the population from calls

    Rangers record the place and time of each call, then compare their observations to avoid counting the same bird several times. The birds sing for about one month each year, during the same period. Two days a week of listening covers only part of that month.

  3. An autonomous acoustic monitoring unit

    Our six-person team built units that attach to trees, recognise capercaillie calls and timestamp detections using GPS. Comparing detections across places and times lets the same cross-check run over the whole singing month, without anyone walking in. I was responsible for the electronics and detection approach. The photograph shows a unit deployed for the field test.

  4. record classify send noise capercaillie 86.6 % of calls · test set 2.1 % false alarms on noise

    Call classification and LoRa transmission

    86.6 % of male calls recognised in testing

    The microphone records sound, an embedded classifier identifies male capercaillie calls, and the device sends detections over LoRa. It does not transmit the audio. On the report’s test set, it recognised 86.6 % of male calls, with a 2.1 % false-positive rate on noise. The report does not document the radio range.

  5. Electronics and energy budget

    The unit contains a controller board, radio and antenna, microphone, GPS, storage and a 1,100 mAh battery. Power drove the choice of platform: a unit that has to be visited to recharge brings people back into the habitat, which is exactly what it was built to avoid.

  6. sleep and recording 30 days required measured · 1,100 mAh 4 days estimated · 10,000 mAh about a month 0102030 days

    Battery-life estimates and prototype measurements

    4 days measured · 30 days required

    The report estimates about one month of operation for a mostly sleeping ESP32 on a 10,000 mAh battery. A Raspberry Pi could run a larger model but would require a solar panel to last a week. The prototype measured 4 days on its 1,100 mAh battery, short of the 30-day requirement.

  7. commercial recorders 150 USD 450 USD 900 USD bill ofmaterials 230 CHF stakeholders ornithologists conservationofficers

    Prototype cost and market comparison

    The project also included a product and market study. The team compared three commercial recorders priced from 150 to 900 USD, estimated a 230 CHF bill of materials for the first prototype, and mapped stakeholders through conversations with conservation officers and ornithologists. The report does not document the field-test duration.

The finished monitoring enclosure strapped to a tree during the field test.
ECO-SENTINEL enclosure mounted on a tree during the field test.
The annotated interior of the enclosure: controller with radio, microphone, GPS, storage and battery.
ECO-SENTINEL electronics inside the enclosure. The platform was selected to meet the power budget.

My contribution

In a six-person product-design team, I was responsible for the electronics and detection approach, including platform selection and power budgeting.