ROBOTIC MOBILITY GROUP MASSACHUSET INSTITUTE OF TECHNOLOGY

Terrain Sensing
 
For mobile robots in rough terrain, the ability to safely traverse terrain is highly dependent on mechanical properties of that terrain. For example, a robot may be able to climb up a rocky slope with ease, but slide down a sandy slope the same grade. With mobile robots being employed for planetary exploration and UGVs being developed for missions on Earth, the ability to predict these mechanical terrain properties from a distance is becoming increasingly important.

This project focuses on:

• Classifying natural terrain based on visual features, such as color, visual texture, and range data,
• Learning visual classification on-line, so that a robot can improve its terrain recognition based on its experiences,
• Autonomously identifying mechanically-distinct terrain classes to eliminate the need for human supervision in establishing the list of terrain classes, and
• Estimating the mechanical terrain properties associated with each of the terrain classes.

The goal of this work is to be able to set a robot down in a previously unexplored environment, and after driving around for a short period of time have it be able to look out in the distance and predict mechanical properties of the terrain it sees.

Experiments for this project have been performed using a four-wheeled mobile robot in natural outdoor terrain. The robot appeared briefly in a segment of NOVA scienceNOW.

This work has been funded by NASA/JPL through the Mars Technology Program. 


Mobility Prediction with Environmental Uncertainty
The ability of autonomous unmanned ground vehicles to rapidly and effectively predict terrain negotiability is a critical requirement for their use on challenging terrain. Most of the work done on mobility prediction for such vehicles, however, assumes precise knowledge about the vehicle/terrain properties. In practical conditions though, uncertainties are associated with the estimation of these parameters. This work focuses on developing efficient methods that take into account environmental uncertainty while determining vehicular mobility.

Omnidirectional Mobile Robots in Rough Terrain Mobile robots are finding increasing use in military, disaster recovery, and exploration applications. These applications frequently require operation in rough, unstructured terrain. Currently, most mobile robots designed for these applications are tracked or Ackermann-steered wheeled vehicles. Methods for controlling these types of robots in both smooth and rough terrain have been well studied. While these robots types can perform well in many scenarios, navigation in cluttered, rocky, or obstacle-dense urban environments can be difficult or impossible. This is partly due to the fact that traditional tracked and wheeled robots must reorient to perform some maneuvers, such as lateral displacement. Omnidirectional mobile robots could potentially navigate faster and more reliably through cluttered urban environments and over rough terrain, due to their ability to track near-arbitrary motion profiles. Currently, the drive mechanisms of most omnidirectional mobile robots are designed to perform well in indoor and benign environments. This project focuses on the analysis, design, and control of omnidirectional mobile robots for use in rough terrain. The robots in this study use active split offset caster drive mechanisms that allow high thrust efficiency during omnidirectional motion and low ground pressures over rough terrain. The design guidelines developed in this research are scalable and applicable for a class of omnidirectional mobile robots. MIT is collaborating with the Illinois Institute of Technology in constructing a prototype robot to experimentally validate the effectiveness of the design guidelines and controller. This work has been funded by the U.S. Army Research Office.

 

Trajectory tracking control for front-steered ground vehicles
The ability to follow a desired trajectory is an important part of many autonomous vehicle navigation and hazard avoidance systems. An important requirement for trajectory tracking controllers is appropriate consideration of the vehicle dynamics, especially with regard to wheel slip. When wheel slip is small, the vehicle dynamics are greatly simplified. When wheel slip does occur, however, it can cause a loss of control, such as in the video below showing a lane change maneuver on snow and ice (please skip to 2 minutes 34 seconds if the video does not automatically do so).
One approach to dealing with the loss of control when wheel slip is large is the use of electronic yaw stability control systems. These systems operate by precisely controlling the brakes at individual wheels to minimize sideslip. Such a system is illustrated in the video above at 3 minutes, 55 seconds. These systems have been shown to reduce the risk of crashes, and fatal crashes in particular.
Electronic stability control systems are effective in reducing wheel slip, but they currently do not consider the effect that stability control has on altering the vehicle path and its ability to avoid collisions with hazards. Also, there is evidence that vehicles can be controlled precisely even with large amounts of wheel slip, as evidenced by the Ken Block Gymkhana video shown below (please skip to 2 minutes, 7 seconds if the video does not automatically do so).
Rather than minimizing wheel slip like a conventional stability controller, we suggest compensating for slip while following a desired trajectory, in a similar manner to expert rally drivers. We recently considered a controller that employs feedback control of tire friction forces to control the position of the front center of oscillation along a desired trajectory. This work was inspired by previous work by Ackermann and is similar to concurrent work being done at Stanford's Dynamic Design Lab.
The trajectory tracking controller is based on a planar half-car model that has one steerable wheel at the front and one wheel at the rear. It is also sometimes called a bicycle model, though it is only 2-dimensional and cannot tip over like a 3-dimensional bicycle. An illustration of this model is given below. Friction force F_f and F_r act at the front and rear wheels, and the speed at the center of gravity (c.g.) is V.

The trajectory tracking controller controls the position of a point near the front wheels to follow a desired trajectory. The vehicle behavior is illustrated in the animation below for a sinusoidal trajectory with low acceleration. It can be seen that the front of the vehicle follows the desired trajectory, and the vehicle orientation oscillates a small amount.