Content-type: text/html Set-Cookie: cookiehash=D8TIX1F9GET8DML97LCWDC1UDL31CF7Q; expires=Tue, 10 Dec 2019 00:00:00 GMT; path=/; DMI News

DMI News

Previous Entry.. Next Entry..


November 17, 2011 19:41

Well, the second critter project is finally underway. I've finally broken down and purchased the necessary component that's been stalling the project. While all of the software and circuitry is simple (if not time consuming and tedius) to construct, when designing the drive system, there's usually always one component that simply doesn't exist. The issue at hand is to find a combination of parts that all work together, the parts being the wheel, motor, and encoder. If a suitable motor can't be found, then you'd also need a drive train consisting of several gears to slow down the motor and increase the torque. For an encoder to work, the motor has to have a secondary shaft to attach it to. And to connect the primary motor shaft to the wheel you need a hub. The hub has been the single component that has held me up, since apparently no such object exists that can connect a 6mm motor shaft to a wheel with a 29/32" bore. To solve this problem, I have used Emachineshop to manufacture the part for me. This is a moderately expensive solution, but the best I could come up with. It will also take a month to deliver, but I've got plenty to do in the meantime.

The new critter is going to have more features available than the critter of the past, or the past rc cars. With an encoder on the motors, I'll be able to track odometry. This will be utilized to keep the robot moving in a straight line, control speed, and track the position relative to a known starting position. Since, due to wheel slip and other factors, odometry is error-prone and that error increases with the distance travelled. For that reason, it is necessary to periodically recompute the current position in the house. Several possibilities for this are possible, many of which involve beacons of some sort. However, I will likely implement a SLAM (Simultanious Localization And Mapping) solution using a sweeping sensor array with combined sonar and IR rangefinders. With a known map either computed or previously provided, the robot will be able to determine its position in the house based on the distance to the objects around it, offset by a predicted area due to odometry measurements.

To solve the battery life problem, in addition to being able to swap out batteries, the critter itself will be able to locate and dock with a charging station whenever it becomes idle or low on battery. A more elaborate set of sensors will be utilized to avoid all contact with walls and other objects. I'm still pondering the prospects of adding on a controllable arm that won't interfere with navigation or weight distribution. With work hopefully slowing down over the next few weeks, I hope to get a lot of the preliminary work done on this project before the part arrives to complete it.