Autonomous Navigation in Complex Environments

TL;DR

This paper presents a CNN-DNN fusion approach for autonomous robot navigation in complex, subterranean environments, trained via imitation learning and tested for robustness in simulation.

Contribution

It introduces a novel fusion of CNN and DNN for navigation control in subterranean rescue scenarios, trained with imitation learning from sensor data.

Findings

Successful navigation in simulated subterranean environments

Model robustness verified through Monte Carlo testing

Effective use of LiDAR and camera data for control

Abstract

This paper explores the application of CNN-DNN network fusion to construct a robot navigation controller within a simulated environment. The simulated environment is constructed to model a subterranean rescue situation, such that an autonomous agent is tasked with finding a goal within an unknown cavernous system. Imitation learning is used to train the control algorithm to use LiDAR and camera data to navigate the space and find the goal. The trained model is then tested for robustness using Monte-Carlo.