Self-Adaptive Autonomous Navigation Based on Reservoir Computing in Snowy Environments
Fangzheng Li, Yonghoon Ji
AI summary
Problem
Autonomous navigation in snowy environments is hindered by obscured terrain, sensor noise, and seasonal changes, while traditional reinforcement learning struggles with the sim-to-real gap and requires extensive simulation training.
Approach
The method combines heuristic reinforcement learning with reservoir computing and artificial bee colony optimization for direct real-world training, alongside a thermal and grayscale-based method to detect and target snow-covered regions.
Key results
- 95.5% pixel accuracy in unsupervised snow-region segmentation
- Direct real-world reinforcement learning training without simulation
- Zero imminent collisions achieved after successive training laps
- Reliable obstacle avoidance and snow targeting on varied real-world terrains
Why it matters
Provides a practical, simulation-free navigation solution for autonomous snow removal robots operating in harsh, dynamically changing winter conditions.
Abstract
Autonomous navigation in snowy environments is essential for snow removal robots operating in regions with heavy snowfall. However, snow accumulation obscures terrain features and introduces sensor noise, making reliable perception and navigation difficult. Moreover, snow removal robots typi- cally operate only during winter, while the environment may change during other seasons, requiring the robot to adapt to new situations. To address these challenges, this study proposes a self-adaptive navigation framework that learns directly in real snowy environments without relying on simulation. The framework integrates reservoir computing (RC), reinforcement learning (RL), and artificial bee colony (ABC) optimization. In addition, a snow-region detection method based on thermal and grayscale images is introduced to guide the robot toward areas requiring snow removal.