Robo-Rats Locomotion: Car-type Drive
Car-type locomotion is very common in the "real world," but not as common in the "robot world." Car-type locomotion (and its cousin, tricycle locomotion) is characterized by a pair of driving wheels and a separate pair of steering wheels (only a single steering wheel in tricycle locomotion):
Although these rear-wheel drive configurations are the most common, there are also front-wheel drive versions as well in which the steering wheel(s) are also the drive wheel(s). An advantage of front-wheel drive is a smaller turning radius. Consider the figure below:
If this were a rear-wheel drive system it is unlikely that such a sharp turn could be accomplished since the majority of the forward force produced by the rear wheels cannot be used for motion (a 90 degree turn of the front wheel would be impossible in a rear-drive system). However, if the front wheel is driven there is no problem with this amount of turn; indeed, even a 90 degree turn of the front wheel could be performed.
The placement of odometry sensors is an issue in any car-type locomotion system where more than one wheel is providing thrust. As in a car, if the rear wheels are driven by a solid axle and a differential is not used the rear wheels will slip when turning because they must travel different distances (the wheel on the outside travels a farther distance that the wheel on the inside). Since any wheel slip will reduce odometry accuracy, placing the shaft encoder on a wheel which does not slip is advantageous.
A car-type drive is one of the simplest locomotion systems in which separate motors control translation and turning (a big advantage compared to the differential drive system). This is why it is popular for human-driven vehicles. However, the simplicity comes at a price: the car-type drive is a non-holonomic actuation system. A non-holonomic system is one in which the actuators do not directly control one or more of the degrees-of-freedom of the system, but instead are coupled such that orientation becomes much more complicated than in a holonomic system. For example, planar motion requires two degrees of freedom: x and y. Any robot architecture that allows for direct and, possibly simultaneous, motion along the x and y axes would be holonomic. Most robots do not have orthogonal actuators, although there are some that do. But any x,y movement can also be expressed in polar coordinates, combinations or rotations and translations. Therefore any robot that can perform arbitrary rotations and translations is also holonomic. However, a car-type drive cannot perform rotations without translating. Thus, the degrees-of-freedom are linked and the system is non-holonomic. The obvious example of this is parallel parking a car. If a car used a synchro drive system (which is holonomic), then the car could position itself next to an empty parking space, rotate its wheels so that the direction of motion is into the parking space, and then translate into the space. Instead, a car must perform many forward/reverse motions in order to accomplish a sideways translation because there is no actuator that can directly perform sideways movement. The result is a system for which path planning is much more difficult.
Below is a photo of the Fiat robot, built in the Dartmouth Robotics Laboratory by Jon Howell. Fiat uses a tricycle locomotion system with front wheel drive:
Two - One for translation and one for rotating the turning wheel(s).
Simplicity - One of the simplest locomotion systems to implement with one caveat: the turning mechanism must be precisely controlled. A small position error in the turning mechanism can cause large odometry errors.
Planning - Planning is difficult because the system is non-holonomic. Note that the difficulty of a non-holonomic system is relative to the environment. On a highway, path planning is easy because the motion is mostly forward with no absolute movement in the direction for which there is no direct actuation (sideways). However, if the environment requires motion in the direction for which there is no direct actuation, path planning is very hard. In the case of the Robo-Rats competition, a car-type drive could be feasible if it had a tight turning radius.
Back to the Robo-Rats Locomotion page
Back to the main Robo-Rats page
Last modified: 04/04/01 22:30