Autonomous snowplow, controls, feedback
With a team objective to “apply hands-on skills and teamwork to find real-world solutions,” Dunwoody College of Technology students and team mentors made an autonomous snowplow for an Institute of Navigation engineering competition, installing a control system for steering/propulsion and creating a feedback system using sensors and optical encoders. The entry won third place in a collegiate Autonomous Snowplow Competition.
Dunwoody’s team had five AAS degree–seeking students and three faculty advisors: James Jorgenson is a student in engineering drafting and design. Michael Carnis, Keith McClelland, Josh Moses, and Nigel Ystenes are automated systems and robotics students. John McShannock, automated systems and robotics instructor, E.J. Daigle, robotics and manufacturing director, and Tim Flugum, physics instructor, served as mentors.
Dunwoody received two riding lawn tractors as a donation from faculty. The goal was to convert one tractor into a snowplow by purchasing a plow blade, installing a control system for steering/propulsion, and creating a feedback system using sensors and optical encoders. The team sought a simple solution using its strengths with machining skills, PLCs, and electronics. The automated plow had to be powerful enough to remove snow using a 1-meter-wide path in one pass, robust enough to survive the Minnesota winter, and intelligent enough to navigate the U-shaped course.
According to the project report, challenges included integration of hardware, navigation sensors, and plow operation.
“The mower we wanted to use had major wire harness problems. Many of the wires were missing, and as we were cutting the steering shaft with a torch we did more damage.” The older tractor also used a threaded lead screw with a collar that rode on the screw, the “preferred method for steering; however, after much work on other systems this tractor was deemed inadequate, and we selected the second tractor. Since we had two tractors, we were able to use the first one as a sacrificial lamb for the purpose of learning.”
“We also had some issues with our first attempt at steering control. At first, we selected a stepper motor to turn the steering shaft. The driver chip we used to control it was unable to handle the current that the motor required. This led us to try switching relays with the driver and then running the stepper with the relays. We prototyped this using mechanical relays but they would not switch fast enough, so we decided to purchase some solid-state relays. The solid-state relays we bought were surplus items, and one was bad when we got it. We replaced the bad relay and then fried another one.
“We gave up on the relays and decided to build our own switching circuit. We used high-power transistors and managed to achieve a top speed of about 24 rpm. We were unable to supply a full 5 V to the stepper so our torque was lower than it should have been. We managed to get 4 V to the stepper…enough to turn our steering shaft with a 5-to-1 gearbox, but it would only turn it in one direction reliably. We then decided to eliminate the stepper and found a 12V dc, 31 rpm motor [from Grainger] that also had sufficient torque to reliably turn the steering shaft.”
“We then attached an encoder wheel to the coupler. The encoder gives us a resolution of 15 degrees per hole for a total of 24 holes equally spaced around the circumference of the encoder wheel. We have also mounted a slotted opto-coupler so that the encoder wheel passes through the slot and pulses the signal from the opto-coupler. We use the signal from the opto-coupler to increment or decrement an up-down counter. When we turn the wheels right, the counter will increment until it reaches a set value. When we are ready to turn the wheels back to straight, the counter will decrement until it reaches 0. We change the direction of rotation by using two separate relays to reverse the polarity of the voltage applied to the power leads on the motor.”
For the programmable logic controller (PLC), “we used a Siemens logo type OBA4 with 2 analog inputs and 6 discrete inputs. We did not consider…that we would not have access to built-in math functions when we selected this PLC. This made it almost impossible to do a comparison between two sources of input, and that meant we could not easily do course corrections.” Inputs I7 and I8 on the PLC also allow the use of analog input (0-10 V dc) for the analog distance sensor. A 4-input expansion module helped drive the tractor from a local joystick for on-site loading, unloading, and setup.
Total hardware project cost was estimated at $3,309; actual cost was $949 thanks to donations from faculty, Banner Engineering, and Cambridge Metals and Plastics. The Dunwoody College of Technology entry placed third competing against five university entries in the 2011 Annual Autonomous Snowplow Competition.
If you like Minnesota in January (or if you like autonomous snow plows even more than you like Minnesota in January), the Second Annual ION Autonomous Snowplow Competition is Jan. 26-29, in St. Paul.
– Edited by Mark T. Hoske, CFE Media, Control Engineering; posted by Chris Vavra, Control Engineering, www.controleng.com