Our preprint on the Terracorder ground sensing platform I've been working with Josh Millar at Imperial on is now available on arXiv. It's a heady combination of ESP32 very low power hardware, combined with Q-learning to build cooperative networks of them that can run for long periods of time without wasting energy on redundant operations. The Terracorder is a versatile multi-sensor device designed for biodiversity monitoring in remote environments. Josh's clever on-device reinforcement learning scheduler captures more than 80% of events at less than 50% of the number of activations of the best-performing fixed schedule. We also explore how a collaborative scheduler can maximize useful operation across a network of devices, improving overall power consumption and robustness - crucial for long-term deployment in places like tropical forests.