Robots for the Forest

Exploring how robotics can assist with forest management and wildfire prevention.

California's wildfire crisis demands new solutions. One approach is to thin out the small diameter material in the forests to reduce catastrophic fire risk. I've been exploring whether robotics could help scale forest thinning operations. After all, is there anything cooler than a robot armed with a chainsaw?

Treespotte 1 robot

Treespotte 1

Built on an RC car chassis, this early attempt faced limitations: restricted mounting space, complicated Ackerman steering, and compatibility issues between Intel RealSense cameras and Jetson Nano hardware. I concluded a fresh start was necessary.

Treespotte 2 robot

Treespotte 2

This version suffered from excessive complexity: three computing systems (Jetson Nano, Latte Panda Alpha, custom Teensy PCB) created networking complications. The robot became too heavy for its differential drive to maneuver effectively. It was overdone.

Treespotte 3 robot

Treespotte 3

A success with simplified design: Latte Panda compute, Intel T265 tracking camera, Lidar, and RoboClaw motor controller. The robot successfully maps indoor spaces and navigates autonomously. I forked an old ROS1 driver for RoboClaw compatibility with ROS2.

Treespotte 4 robot

Treespotte 4

Current focus targets outdoor autonomous navigation and object recognition (pine cones, trees). Uses ZED 2i stereo depth camera, GPS integration, and Jetson AGX Orin Developer Kit. Chassis completed April 2022 using 3D-printed PLA parts designed in Fusion 360.

Resources