What question did this study set out to answer?

April 1, 2026Open Access

Teaching a Real Biped to Walk with Neuro-Evolution After Making Tests and Comparisons on Simulated 2D Walkers

RSRoland SzabóPolytechnic University of Timişoara

Key Points

The aim is to evaluate different neuro-evolution methods to train a simulated biped walker and apply the best one to a real biped.
Tested four neuro-evolution techniques: DQN, NEAT, DDPG, ARS.
Conducted simulated walking trials to assess each method's effectiveness.
Ported the best method to a real biped for practical training.
ARS showed the best simulation results for training the walker.
NEAT was the easiest and fastest method for use on the actual biped, despite ARS's superior simulation performance.

Abstract

The aim of this paper is to test and compare different neuro-evolution methods to train a simulated biped walker to learn to walk. After this step, the best neuro-evolution technique is ported to a real biped, which would train itself to walk. The goal is to reduce the number of falls for the real biped in order to avoid destroying the physical unit. The following four neuro-evolution methods were tested: Deep Q-Learning (DQN), NeuroEvolution of Augmenting Topologies (NEAT), Deep Deterministic Policy Gradients (DDPG), and Augmented Random Search (ARS). The best results from simulations were obtained with the ARS method, but the fastest and easiest to implement on the real biped was the NEAT algorithm.

Ask AI

Helpful

Bookmark

View Full Paper