Mantegh, I., Jenkin, M. R. M., and Goldenberg, A. A. Reformulating the potential field method for goal-attaining, real-time path planning. Proc. 3rd ECPD Int. Conf. on Advanced Robotics, Intelligent Automation and Active Systems, Bremen, Germany, 132-136, 1997.
The objective of path planning is to find a sequence of states that a system has to visit in order to attain the goal state. Because of their real-time efficiency, potential field methods present a powerful heuristic to guide this search. However, potential field approaches can not guarantee goal attainability. They are often referred to as "local methods" and are used in conjunction with a global path planning algorithm. The present work introduces a novel methodology for path planning which combines the real-time efficiency of potential field methods with goal attainability characterisitcs of global methods (such as A*). The algorithm of this work is: i) free from local minima; ii) capable of considering arbritary-shaped obstacles (no geometric approximation is required); iii) computationally less complex than previous obstacle avoidance and goal attainability at the same time.
Dudek, G., Jenkin, M., Milios, E., and Wilkes, D., Map validation and self-location for a robot with a graph-like map, Robotics and Autonomous Systems, 22:159-178, 1997.
This paper deals with the validation of topological maps of an environment by an active agent (such as a mobile robot), and the localization of an agent in a given map. The agent is assumed to have neither compass nor other instruments for measuring orientation or distance, and therefore, no associated metrics. The topological maps considered are similar to conventional graphs. The robot is assumed to have enough sensory capability to transverse graph edges autonomously, recognize when it has reached a vertex, and enumerate edges incident upon the current vertex, locating them relative to the edge via which it entered the current vertex. In addition, the robot has access to a set of visually detectable, portable, distinct markers. We present algorithms, along with worst case complexity bounds and experimental results for representative classes of graphs for two fundamental problems. The first problem is the validation problem: if the robot is given an input map and its current position and orientation with respect to the map, determine whether the map is correct. The second problem is the self-location problem: given only a map of the environment, determine the position of the robot and its "orientation" (i.e., the correspondence between edges of the map and edges in the world at the robot's position). Finally, we consider the power of some other non-metric aids in exploration.
Jenkin, M., and Harris, L., (Eds.) Computational and Psychophysical Mechanisms of Visual Coding, Cambridge University Press, 1997.
All visual tasks, from the simplest computer graphics program to the most complex biological visual system require an underlying representation of visual information. The structure or coding of this representation provides the framework for processing the information. Both the biological and computational communities have had to address the task of designing or inferring visual coding strategies. This volume, with chapters by some of the most active contributors in the field of visual coding, illustrates the similarities in the problems considered and the common models and algorithms that are proposed to solve them. Researchers in neuroscience and computational vision will find a wealth of new ideas here.
