Building Forest Inventories with Autonomous Legged Robots -- System, Lessons, and Challenges Ahead
Matias Mattamala, Nived Chebrolu, Jonas Frey, Leonard FreiÃmuth, Haedam Oh, Benoit Casseau, Marco Hutter, Maurice Fallon
AI summary
Problem
Ground-level forest mapping is labor-intensive and currently relies on manual methods or limited aerial platforms that struggle under dense canopies. There is a need for robust, autonomous ground platforms that can navigate unstructured natural environments to collect fine-scale tree data.
Approach
The authors developed an integrated autonomy system for a quadruped robot that combines LiDAR-inertial odometry, SLAM, dense mapping, and online tree detection to autonomously navigate forests and estimate tree traits in real time.
Key results
- Surveyed up to 1-hectare forest plots autonomously in under 30 minutes
- Achieved 2 cm accuracy in estimating tree diameter at breast height
- Validated system performance across five field campaigns in three European countries
- Identified five key lessons and challenges for legged robot navigation in unstructured forests
Why it matters
Provides foresters and roboticists with a scalable, low-impact alternative to manual ground surveys, advancing the adoption of legged robots in environmental monitoring and sustainable forestry.
Abstract
Legged robots are increasingly being adopted in industries such as oil, gas, mining, nuclear, and agriculture. However, new challenges exist when moving into natural, less-structured environments, such as forestry applications. This article presents a prototype system for autonomous, undercanopy forest inventory with legged platforms. Motivated by the robustness and mobility of modern legged robots, we introduce a system architecture, which enabled a quadruped platform to autonomously navigate and map forest plots. Our solution involves a complete navigation stack for state estimation, mission planning, and tree detection and trait estimation. We report the performance of the system from trials executed over one and a half years in forests in three European countries. Our results with the ANYmal robot demonstrate that we can survey plots up to 1-ha plot under 30 min while also identifying trees with typical diameter at breast height (DBH) accuracy of 2 cm. The findings of this project are presented as five lessons and challenges. In particular, we discuss the maturity of hardware development, state estimation limitations, open problems in forest navigation, future avenues for robotic forest inventory, and more general challenges to assess autonomous systems. By sharing these lessons and challenges, we offer insight and new directions for future research on legged robots, navigation systems, and applications in natural environments. Additional videos can be found in https://dynamic.robots.ox.ac.uk/projects/legged-robots