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

Daniel Fuertes, Andrea Cavallaro, Carlos R del-Blanco, Fernando Jaureguizar, Narciso García

Primary Area: applications to robotics, autonomy, planning

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: deep reinforcement learning, transformer, path planning, route planning, navigation

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

TL;DR: An approach to the navigation problem, which combines high-level route planning (waypoint sequencing) with low-level path planning (collision-free trajectory prediction), using a Transformer-based arquitecture trained with deep reinforcement learning

Abstract: Automatic path planning is a highly relevant research area with multiple applications, but it 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 more 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 high competitive accuracy from NaviFormer since it can understand the constraints and difficulties of each high- and low-level planning and act consequently to improve the performance. Moreover, its superior computation speed proves its suitability for real-time applications.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 5627