Vector Field Driven Design for Lightweight Signal Processing and Control Schemes for Autonomous Robotic Navigation
- PDF / 1,173,127 Bytes
- 9 Pages / 600.05 x 792 pts Page_size
- 65 Downloads / 166 Views
Research Article Vector Field Driven Design for Lightweight Signal Processing and Control Schemes for Autonomous Robotic Navigation Nebu John Mathai, Takis Zourntos, and Deepa Kundur Department of Electrical and Computer Engineering, Texas A&M University, College Station, TX 77840, USA Correspondence should be addressed to Nebu John Mathai, [email protected] Received 31 July 2008; Revised 26 February 2009; Accepted 8 April 2009 Recommended by Frank Ehlers We address the problem of realizing lightweight signal processing and control architectures for agents in multirobot systems. Motivated by the promising results of neuromorphic engineering which suggest the efficacy of analog as an implementation substrate for computation, we present the design of an analog-amenable signal processing scheme. We use control and dynamical systems theory both as a description language and as a synthesis toolset to rigorously develop our computational machinery; these mechanisms are mated with structural insights from behavior-based robotics to compose overall algorithmic architectures. Our perspective is that robotic behaviors consist of actions taken by an agent to cause its sensory perception of the environment to evolve in a desired manner. To provide an intuitive aid for designing these behavioral primitives we present a novel visual tool, inspired vector field design, that helps the designer to exploit the dynamics of the environment. We present simulation results and animation videos to demonstrate the signal processing and control architecture in action. Copyright © 2009 Nebu John Mathai et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
1. Introduction The problem of developing a control architecture for autonomous robotic agents involves numerous challenges pertaining to the best use of limited, nonideal information. Beyond this, given the remote, energy-scarce environments that robots have found application (e.g., space robotics, underwater exploration, mobile sensor networks deployed in inhospitable, unknown terrain) and the multiagent robotic paradigm, the need for signal processing with lightweight implementation (in terms of area and power complexity, and low-latency autonomous computation) has become increasingly important. To minimize the economic cost of a multiagent system, it is important that the complexity of each agent be constrained. Moreover, in robotic exploration problems (where the agent must be able to maneuver effectively through challenging and inaccessible environments) and mobile sensor network applications, low agent complexity (e.g., in terms of compactness and energy usage) is demanded. Further, it has been suggested [1] that robotics, the endeavor of synthesizing artificial goal-directed machines, may offer insight to biology, the study of goal-directed organisms in
nature. To that end, the development of synthesis methods for autonom
Data Loading...
