# Learning to Navigate in a VUCA Environment: Hierarchical Multi-expert Approach

**Authors:** Wenqi Zhang, Kai Zhao, Peng Li, Xiao Zhu, Faping Ye, Weijie Jiang, Huiqiao Fu, Tao Wang  
**arXiv:** [2111.08364](https://arxiv.org/abs/2111.08364)  
**Date:** 2021-11-16  
**Summary page:** https://www.summarizepaper.com/en/paper/2111.08364v1/

## Key points

- Robot navigation in a real-world scenario with VUCA is challenging despite decades of efforts
- The authors propose a hierarchical multi-expert learning framework for autonomous navigation inspired by the CNS
- The upper layer performs simultaneous map exploration and route-planning to avoid trapping in a blind alley using a heuristic exploration mechanism that considers target location, path cost, and safety level
- The lower layer uses a local adaptive model that fuses multiple discrepant strategies to pursue a balance between collision-avoidance and go-straight strategies while acting as the cerebellum in the CNS
- The proposed algorithm was tested through simulation and real-world experiments on multiple platforms including legged and wheeled robots
- Experimental results demonstrate that it outperforms existing methods in terms of task achievement, time efficiency, and security.

## Summary

Robot navigation in a real-world scenario with volatility, uncertainty, complexity, and ambiguity (VUCA) remains a challenging topic despite decades of efforts. To address this issue, the authors propose a hierarchical multi-expert learning framework for autonomous navigation inspired by the central nervous system (CNS). The upper layer of the proposed framework performs simultaneous map exploration and route-planning to avoid trapping in a blind alley using a heuristic exploration mechanism that considers target location, path cost, and safety level. This layer acts similarly to the cerebrum in the CNS. The lower layer uses a local adaptive model that fuses multiple discrepant strategies to pursue a balance between collision-avoidance and go-straight strategies while acting as the cerebellum in the CNS. The proposed algorithm was tested through simulation and real-world experiments on multiple platforms including legged and wheeled robots. Experimental results demonstrate that it outperforms existing methods in terms of task achievement, time efficiency, and security. The paper is authored by Wenqi Zhang, Kai Zhao, Peng Li, Xiao Zhu, Faping Ye, Weijie Jiang, Huiqiao Fu, and Tao Wang. Their work titled "Learning to Navigate in a VUCA Environment: Hierarchical Multi-expert Approach" was presented at the 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).