NaviFormer: A Deep Reinforcement Learning Transformer-like Model to Holistically Solve the Navigation Problem 

Description

Path planning is usually solved by addressing either the (high-level) route planning problem (waypoint sequencing to achieve the final goal) or the (low-level) path planning problem (trajectory prediction between two waypoints avoiding collisions). However, real-world problems usually require simultaneous solutions to the route and path planning subproblems with a holistic and efficient approach. In this paper, we introduce NaviFormer, a deep reinforcement learning model based on a Transformer architecture that solves the global navigation problem by predicting both high-level routes and low-level trajectories. To evaluate NaviFormer, several experiments have been conducted, including comparisons with other algorithms. Results show competitive accuracy from NaviFormer since it can understand the constraints and difficulties of each subproblem and act consequently to improve performance. Moreover, its superior computation speed proves its suitability for real-time missions.

 For questions about this multi-agent system, please contact Daniel Fuertes at This email address is being protected from spambots. You need JavaScript enabled to view it..

Download

Click here(https://github.com/danifuertes/naviformer) to download the code.

Citation

D. Fuertes, A. Cavallaro, C. R. del Blanco, F. Jaureguizar, N. García, NaviFormer: A Deep Reinforcement Learning Transformer-like Model to Holistically Solve the Navigation Problem, accepted in IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2025.