Robust Mobility and Manipulation for Flexible and Modular Robot Assistants

This thesis examines the design and implementation of flexible and modular robot assistants, with a focus on enhancing mobility and manipulation capabilities for a tracked rescue robot. The research integrates advanced robot modeling, scene understanding, and assistive behaviors to create a versatile robotic platform capable of operating in unpredictable and hazardous conditions.
The developed tracked rescue robot, the focus of this study, is designed for high performance in rough terrain and utilizes advanced mobility algorithms to ensure stability and maneuverability. Through comprehensive robot modeling, the system is optimized to navigate challenging environments while maintaining precise control over its manipulation tasks. This capability is crucial in search and rescue operations, where the robot must navigate through debris, ascend steep inclines, and precisely inspect or manipulate objects.

The need for adaptable and resilient robotic systems is crucial in disaster relief operations. This thesis presents an adaptable robot and sensor payload approach to develop flexible and modular robot assistants for robust mobility and manipulation in dynamic and challenging environments. The focus was on developing and using a lightweight tracked rescue robot characterized by high mobility in rough terrain and precise manipulation capabilities. The robot's design integrates modular components that enable rapid reconfiguration to suit various tasks, enhancing its versatility in unpredictable environments.
The system leverages sophisticated robot modeling techniques to optimize the robot's performance in complex terrains, ensuring stability and maneuverability. Furthermore, we employ state-of-the-art scene understanding algorithms to provide the robot with real-time perception and decision-making capabilities.

Scene understanding is pivotal in the robot's ability to operate safely. The robot can analyze and interpret its surroundings with advanced perception techniques in real-time to inform the end-user of any robot decisions, avoid obstacles, or prioritize tasks. This allows the robot to autonomously assess and navigate hazardous environments, identify key objects, and perform critical rescue operations.
This environmental awareness is complemented by assistive behaviors designed to support human operators in high-stress situations. These behaviors include autonomous task execution, real-time sensor feedback, and adaptive learning mechanisms that enable the robot to refine its operations based on environmental feedback and operator input.

The experimental results demonstrate the effectiveness of the proposed system in simulated disaster scenarios, highlighting its potential as a reliable tool in search and rescue missions. The robot's modularity also offers significant advantages in terms of scalability and customization, making it a versatile platform for a wide range of applications beyond disaster response, including industrial automation and healthcare assistance. The results underscore the potential of this research to make significant contributions to disaster and assistance robotics.