Resilient forest autonomy
This project investigates resilient autonomy for mobile robots operating in complex forest environments. Forests provide a demanding but accessible real-world testbed in which autonomous systems must cope with irregular terrain, dense vegetation, uncertain perception and physical interactions that are difficult to reproduce in controlled laboratory environments.
The research examines how wheeled and legged robots can perceive, navigate and remain operational under these conditions. Current work includes robust navigation with limited or uncertain sensing, detection and characterisation of robot entrapment, and perception-driven approaches to autonomous operation and recovery in forests. A particular emphasis is placed on low-cost and resource-efficient autonomy, reducing dependence on expensive sensing, computation and highly engineered platforms.
The broader aim is to develop autonomous systems that remain effective when the environmental and physical conditions encountered in deployment differ from those assumed during their design.
People
Themes
Publications
- Blind-Wayfarer: A minimalist, probing-driven framework for resilient navigation in perception-degraded environments
- An embarrassingly simple approach for visual navigation of forest environments
- End-to-End learning for visual navigation of forest environments
- Low-viewpoint forest depth dataset for sparse rover swarms