This paper describes the motivation, design and implementation of a Perception Validation System (PVS), a system for measuring the outdoor perception performance of an autonomous vehicle. The PVS relies on using large amounts of real world data and ground truth information to quantify performance aspects such as the rate of false positive or false negative detections of an obstacle detection system. Our system relies on a relational database infrastructure to achieve a high degree of flexibility in the type of analyses it can support. We discuss the main steps required for going from raw data to numerical estimates describing the performance of the perception system, including the generation of ground truth information and the safe speed metric we found to be most useful for comparing the perception system's outputs to the ground truth data. We present results illustrating some of the analyses that can be completed using the Perception Validation System.

Recent evidence that water flowed down the faces of some Martian cliffs has created an interest in the scientific community to explore these cliffs more thoroughly, perhaps using robots. This paper presents a technique for autonomous exploration that considers multiple forms of information while exploring. The method is applied to the problem of autonomously exploring cliffs. The exploring robot seeks out unknown terrain and identifies any cliffs in the environment. When a cliff is found, the robot finds a way to the bottom to view the face of the cliff where signs of water runoff may be present. Results from experiments both in simulation and in the field are presented.

Enables robot explorers to maximize the total information gained while minimizing costs such as driving, sensing and planning. The paper presents a general methodology for solving complex exploration tasks which employs multiple sources of information. The paper also develops a specific instantiation of the method to solve the exploration problem of creating a complete traversability map of an known region. Simulation results showing the solution of this exploration task are included.

The ability to control a robot in both constrained and unconstrained situations is important. One way to do this is using a switching controller which uses a force controller for constrained space and a position controller for unconstrained space. This paper develops a switching controller for a single flexible link manipulator and provides experimental results for a motion from unconstrained to constrained space. Tarn et al. (1996) proposed that positive acceleration feedback be used to reduce the impulse force of contact. The effect of different levels of positive acceleration feedback on the force response, after a collision with an unexpected object occurs, is examined. It is concluded that small amounts of positive acceleration feedback reduces the force overshoot caused by the collision but that too much can cause instability.

A simple collision detection method for a flexible link manipulator is proposed. The method only requires two strain measurements to determine the position of a collision with the flexible link. Equations which relate the force and position of contact to the strain at two points in the flexible link are derived assuming the link is at static equilibrium. Since the flexible robot is actually in motion, one needs to examine whether these relationships are accurate enough in a dynamic scenario to be useful. Thus the expressions are verified on a robot through collision tests, and used to successfully locate an encountered object so that a path can be planned around it.