Improving Path Planning Performance through Multimodal Generative Models with Local Critics

Abstract--This paper presents a novel method for accelerating path planning tasks in unknown scenes with obstacles by utilizing Wasserstein Generative Adversarial Networks (WGANs) with Gradient Penalty (GP) to approximate the distribution of the free conditioned configuration space. Our proposed approach involves conditioning the WGAN-GP with a Variational Auto-Encoder in a continuous latent space to handle multimodal datasets. However, training a Variational Auto-Encoder with WGAN-GP can be challenging for image-to-configuration-space problems, as the Kullback-Leibler loss function often converges to a random distribution. To overcome this issue, we simplify the configuration space as a set of Gaussian distributions and divide the dataset into several local models. This enables us to not only learn the model but also speed up its convergence. We evaluate the (a) Approximation of the freesearch (b) Approximation of the freesearch reconstructed configuration space using the homology rank of tree of a path with tree of a path with manifolds for datasets with the geometry score. Furthermore, we WGAN-GP and a VAE for an MultiWGAN-GP for an unseen propose a novel transformation of the robot's configuration space unseen scenario scenario space that enables us to measure how well collision-free regions are Our experiments show promising results for accelerating path planning tasks in unknown scenes while generating quasioptimal paths with our WGAN-GP. However, these This work utilizes an RGB image of the robot's imagescenario algorithms can be time-consuming, as they involve exploring to encode the input condition. The aim is to convert the configuration space randomly. This becomes even more discrete training data into a continuous representation, which challenging when the CS is complex and difficult to model empowers the generator to create unconstrained configuration analytically, which is often the case in practical applications.