Unique Approaches to Utilization of Proprioceptive and Exteroceptive Sensor Systems for Autonomous Agile Mobility

Principal Investigator

• Mohammad Haider {PI}, Vladimir Vantsevich, Samuel Misko (Co‐PIs), University of Alabama, Birmingham

Student

• Moataz Khalifa (Postdoc then Research Scientist), Steven Gardner, University of Alabama, Birmingham

Government

• David Gorsich, Jon Smereka, U.S. Army GVSC

Industry

• Torsten Kluge, dSPACE GmbH

Project #1.A83 began mid-2020 and ended Q1-2022.

This project seeks to develop and define new fundamental approaches and provide recommendations for the utilization of proprioceptive and exteroceptive sensor systems for autonomous agile mobility.

The maximization of vehicle mobility in off-road environments requires high speed iterative intervention by a vehicle’s mobility control systems during the evolution of a critical motion situation. This stands in contrast to existing modern electronic vehicle control systems such as traction control, wheel torque vectoring systems, and anti-lock braking which are largely designed to take affect after the vehicle has begun to experience loss in mobility due to a critical motion situation.

To make the paradigm shift from reaction control to agile control of mobility, fundamental research to define, characterize, and evaluate various configurations of both available and theoretical sensors, implementation details, cabling, data acquisition, and post-processing solutions. Modeling, simulation, evaluation, and exploration of these solutions (along with their realistic error sources) provide a unique contribution by providing a common reference point and practical evaluation methodology for vehicle system and sensor designers of the future.

The recommendations and guidelines that result from this multi-year effort establishes performance requirements, recommendations, and performance thresholds for each critical component in the sensor system, as well as a methodology for estimating their combined effect on wheel speed and mobility estimation.

A Modified Echo State Network for Fast Semantic Segmentation of Unregulated Terrains was developed.

Other Publication:

• Gardner, S., Haider, M. R., Smereka, J., Jayakumar, P., Gorsich, D., Moradi, L., & Vantsevich, V. (2021). Rapid High-Dimensional Semantic Segmentation with Echo State Networks. Ground Vehicle Systems Engineering Technology Symposium (GVSETS) - Autonomy, Artificial Intelligence, Robotics (AAIR).

1.A83