Autonomous robot control: integrated control strategies for a mobile robotic arm

Abstract

Autonomous robots open up a wide range of potential applications for robotic systems beyond the controlled manufacturing environments where they were originally used. These applications can include agriculture, space exploration, search and rescue, fire-fighting, performing tasks in hazardous environments, personal care robots, and much more. Some of these applications have the potential to replace humans in dangerous environments or improve quality of life for elderly or disabled individuals, thus providing great positive societal impacts. However, the technologies needed for robots to operate safely and autonomously in unstructured environments, and especially when interacting with humans, are still being developed. Designing and controlling autonomous robotic systems is a very challenging problem, with some of the major objectives including efficient autonomous navigation in both known and unknown environments; real-time, dynamic obstacle avoidance; real-time and energy-efficient trajectory generation; and safe operation with and/or in close proximity to humans. While all of these topics have been researched in the field of robotics, existing solutions still have limitations which encourage further developments improving on existing autonomous robotic capabilities. Furthermore, each application and configuration of autonomous robotic systems has a unique set of requirements. In this dissertation, the platform of a mobile robotic arm was chosen for its wide range of potential applications achieved from the combined navigation and manipulation abilities of such a robot configuration. Within the scope of autonomous mobile robot arm control, the following topics were identified and chosen for research: (1) indoor target localization; (2) efficient navigation in partially known environments; (3) integrated control (i.e. coordinated base and arm motion) of a mobile robot arm, including both navigation and trajectory generation and tracking. For each topic, a novel or improved methodology was developed, all relevant to a wide range of autonomous robot deployments. The contributions of this dissertation are as follows: (1) Bluetooth-based homing controller for indoor target localization achieving a mean target localization accuracy of 0.12m or less with various levels of simulated sensor noise; (2) modified artificial potential field-based method for efficient navigation in a partially-known environment and for integrated control of a mobile robot arm improving autonomous navigation success rate and efficiency over existing APF-based methods in environments of varying levels of complexity; (3) real-time many objective optimization-based approach trajectory generation method for integrated motion of a mobile robot arm to reach a desired end-effector configuration, demonstrating a 100% success rate in achieving the desired configuration and reaching the configuration in under 30s in 77% of trials; (4) offline trajectory generation for mobile robot arm end-effector trajectory tracking using a sampling-based combinatorial optimization method to generate integrated motion trajectories (coordinated mobile base and robotic arm motion) achieving over 99% success rate in high accuracy (<5mm position tracking error and <0.1 radian orientation tracking error) end-effector trajectory tracking tested on 500 sample trajectories; and (5) integrated controller design for differential drive mobile robot arm trajectory tracking consisting of a fuzzy logic-based differential drive robot (DDR) controller reducing irregular trajectory tracking errors by 2.4X to 6.8X over existing DDR controller designs, and integrated robotic arm-facilitated DDR base tracking error compensation reducing mean maximum end-effector tracking errors by 18%.