International Journal of Innovative Research in Computer and Communication Engineering
ISSN Approved Journal | Impact factor: 8.771 | ESTD: 2013 | Follows UGC CARE Journal Norms and Guidelines
| Monthly, Peer-Reviewed, Refereed, Scholarly, Multidisciplinary and Open Access Journal | High Impact Factor 8.771 (Calculated by Google Scholar and Semantic Scholar | AI-Powered Research Tool | Indexing in all Major Database & Metadata, Citation Generator | Digital Object Identifier (DOI) |
| TITLE | Implementation of a NEAT-Based Reinforcement Learning System for Autonomous Vehicle Navigation on ESP32 |
|---|---|
| ABSTRACT | This paper presents the complete design and implementation of an autonomous vehicle navigation system that unifies NeuroEvolution of Augmenting Topologies (NEAT) with a dense Reinforcement Learning (RL) reward signal to evolve controllers capable of collision-free path following in complex, obstacle-laden environments. NEAT evolves neural network topologies — adding and removing nodes and connections — across 40 generations of 50 genomes, using cumulative RL reward as the fitness function. A multi-objective A* planner with a five-stage path optimisation pipeline (RDP simplification, line-of-sight pruning, obstacle pushing, collision validation) supplies the reference path, and a 30+ terrain type, 5-wheel-profile vehicle model ensures physical plausibility. The best-evolved genome is deployed to an ESP32 microcontroller over Wi-Fi, which executes move/turn commands via a PID-regulated L293D motor driver with quadrature encoder feedback. Experimental results demonstrate convergence at generation 30, a 71% reduction in collisions compared to A*-only planning, a 94% simulation success rate, and a 10 Hz real-time execution rate on hardware. |
| AUTHOR | PROF. VAISHNAVI SONAWANE, MASOOD MADKI Diploma Student, Department of Artificial Intelligence & Machine Learning, AISSMS Polytechnic, Pune, India |
| VOLUME | 182 |
| DOI | DOI: 10.15680/IJIRCCE. 2026.1403073 |
| pdf/73_Implementation of a NEAT-Based Reinforcement Learning System for Autonomous Vehicle Navigation on ESP32.pdf | |
| KEYWORDS | |
| References | [1] K. O. Stanley and R. Miikkulainen, Evolving neural networks through augmenting topologies, Evol. Comput., vol. 10, no. 2, pp. 99-127, 2002. [2] R. S. Sutton and A. G. Barto, Reinforcement Learning: An Introduction, 2nd ed., MIT Press, 2018. [3] A. Y. Ng, D. Harada, and S. Russell, Policy invariance under reward transformations: Theory and application to reward shaping, ICML, 1999. [4] P. E. Hart, N. J. Nilsson, and B. Raphael, A formal basis for the heuristic determination of minimum cost paths, IEEE Trans. Syst. Sci. Cybern., vol. 4, no. 2, pp. 100-107, 1968. [5] M. Likhachev, G. Gordon, and S. Thrun, ARA*: Anytime A* with provable bounds on sub-optimality, Adv. Neural Inf. Process. Syst., 2003. [6] K. J. Astrom and T. Hagglund, PID Controllers: Theory, Design, and Tuning, ISA Press, 1995. [7] Texas Instruments, L293x Quadruple Half-H Drivers, Datasheet SLRS008H, 2022. [8] Espressif Systems, ESP32 Technical Reference Manual v5.1, 2023. [9] D. H. Douglas and T. K. Peucker, Algorithms for the reduction of the number of points required to represent a digitized line, Cartographica, vol. 10, no. 2, pp. 112-122, 1973. [10] J. Togelius et al., Search-based procedural content generation: A taxonomy and survey, IEEE Trans. Comput. Intell. AI Games, 2011. [11] N. Jakobi, P. Husbands, and I. Harvey, Noise and the reality gap: The use of simulation in evolutionary robotics, ECAL, 1995. [12] S. Thrun, W. Burgard, and D. Fox, Probabilistic Robotics, MIT Press, 2005. |
