Abstract. Autonomous navigation in cross-country environments presents many new challenges with respect to more traditional, urban environments. The lack of highly structured components in the scene complicates the design of even basic functionalities such as obstacle detection. In addition to the geometric description of the scene, terrain typing is also an important component of the perceptual system. Recognizing the different classes of terrain and obstacles enables the path planner to choose the most efficient route toward the desired goal.This paper presents new sensor processing algorithms that are suitable for cross-country autonomous navigation. We consider two sensor systems that complement each other in an ideal sensor suite: a color stereo camera, and a single axis ladar. We propose an obstacle detection technique, based on stereo range measurements, that does not rely on typical structural assumption on the scene (such as the presence of a visible ground plane); a color-based classification system to label the detected obstacles according to a set of terrain classes; and an algorithm for the analysis of ladar data that allows one to discriminate between grass and obstacles (such as tree trunks or rocks), even when such obstacles are partially hidden in the grass. These algorithms have been developed and implemented by the Jet Propulsion Laboratory (JPL) as part of its involvement in a number of projects sponsored by the US Department of Defense, and have enabled safe autonomous navigation in high-vegetated, off-road terrain.

[1] The cratered plains of Gusev traversed by Spirit are generally low-relief rocky plains dominated by impact and eolian processes. Ubiquitous shallow, soil-filled, circular depressions, called hollows, are modified impact craters. Rocks are dark, fine-grained basalts, and the upper 10 m of the cratered plains appears to be an impact-generated regolith developed over intact basalt flows. Systematic field observations across the cratered plains identified vesicular clasts and rare scoria similar to original lava flow tops, consistent with an upper inflated surface of lava flows with adjacent collapse depressions. Crater and hollow morphometry are consistent with most being secondaries. The sizefrequency distribution of rocks >0.1 m diameter generally follows exponential functions similar to other landing sites for total rock abundances of 5-35%. Systematic clast counts show that areas with higher rock abundance and more large rocks have higher thermal inertia. Plains with lower thermal inertia have fewer rocks and substantially more pebbles that are well sorted and evenly spaced, similar to a desert pavement or lag. Eolian bed forms (ripples and wind tails) have coarse surface lags, and many are dust covered and thus likely inactive. Deflation of the surface $5-25 cm likely exposed two-toned rocks and elevated ventifacts and transported fines into craters creating the hollows. This observed redistribution yields extremely slow average erosion rates of $0.03 nm/yr and argues for very little long-term net change of the surface and a dry and desiccating environment similar to today's since the Hesperian (or $3 Ga).

Abstract-The goal of the Conro Project is to build deployable modular robots that can reconfigure into different shapes such as snakes or hexapods. Each Conro module is, itself, a robot and hence a Conro robot is actually a multirobot system. In this paper we present an overview of the Conro modules, the design approach, an overview of the mechanical and electrical systems and a discussion on size versus power requirement of the module. Each module is self-contained; it has its own processor, power supply, communication system, sensors and actuators. The modules, although self-contained, were designed to work in groups, as part of a large modular robot. We conclude the paper by describing some of the robots that we have built using the Conro modules and describing the miniature custom-made Conro camera as an example of the type of sensors that can be carried as payload by these robots.

In this paper we introduce visual compliance, a new vision-based control scheme that lends itself to task-level speci cation of manipulation goals. Visual compliance is e ected by a hybrid vision/position control structure. Speci cally, the two degrees of freedom parallel to the image plane of a supervisory camera are controlled using visual feedback, and the remaining degree of freedom (perpendicular to the camera image plane) is controlled using position feedback provided by the robot joint encoders. With visual compliance, the motion of the end e ector is constrained so that the tool center of the end e ector maintains \contact" with a speci ed projection ray of the imaging system. This type of constrained motion can be exploited for grasping, parts mating, and assembly.We begin by deriving the projection equations for the vision system. We then derive equations used to position the manipulator prior to the execution of visual compliant motion. Following this, we derive the hybrid Jacobian matrix that is used to e ect visual compliance. Experimental results are given for a number of scenarios, including grasping using visual compliance.