System Implementation

Frame

The chassis or frame subsystem shown in Figure 7.1.1 is designed with robustness and reliability in mind. It is composed of 80/20 aluminum extrusion with a cross section of 20x20mm. M5 sliders are used to fasten components to the extrusion bars and to fasten the bars to each other. An 80/20 frame results in a sturdy chassis which in turn optimizes the reliability of the subsystems (drivetrain, front lift, and rear lift mechanisms) which attach onto the chassis.

Drive Train

The drivetrain includes 4 dc motors and 4 mecanum wheels. This setup allows for full omnidirectional motion. In turn, the chassis’s size is optimized to be as large as possible to cover the stair without hindering its ability to maneuver effectively. Drive motors were selected with a torque/speed curve in mind to move 2.5lbs/motor across the step in 15 seconds.

Lift Mechanism

Sensing

We originally planned to include a wide array of sensors around the robot to feed key environmental information back to the Arduino. This data would be used by our control algorithms to dictate the robot's motion, maneuvering, and climbing. We initially decided on Ultrasonic Distance Sensors as the ideal sensor type since they seemed accurate and easy to integrate in the Sensors Lab. The HC-SR04 Ultrasonic Distance Sensor is an extremely cheap and commonly-used sensor that provides reasonable measurement accuracy within a meter. We later discovered that these sensors lacked precision when the distance to a target is less than 5cm. We also ran into an issue with sensor cross-talk, where multiple ultrasonic distance sensors in close proximity to each other would pick up the signals from each other, leading to faulty measurements. We fixed the latter problem by staggering our sensor signals to prevent cross-talk. However, in order to compensate for the lackluster close-range performance of the ultrasonic distance sensors, we employed an additional sensor type. We debated using a LIDAR, IR distance sensors, and push buttons before settling on Limit Switches. These were mounted on either side of the robot as well as on the bottom of the main frame near the rear lift. These allowed us to determine when the robot was in contact with a wall or when the robot's main body was in contact with a stair. Our original plan was to include 7 limit switches: 2 for the front, 2 for the left side, 2 for the right side, and one pointing downward. We eventually cut this number down to 4 (2 front, 1 left, 1 right) before finally replacing the side ultrasonic distance sensors with limit switches. This left us with two ultrasonic distance sensors at the front of the robot, though our robot could accommodate an additional two (one on each side). Removing redundant sensors allowed us to maximize our limited number of GPIO pins on our Arduino Mega. However, redundant sensors could have proven useful by allowing us to compare the measurements and "sanity check" incoming data signals. Two front sensors allowed the robot to compare the measurements from each sensor and pivot to keep the measurements roughly the same. This aligned the robot parallel to the step so it could move along its length. In practice, this was a complex and sometimes unsuccessful maneuver.

The HC-SR04 Ultrasonic Distance Sensors and the Limit Switches were mounted in custom 3D-printed PLA housings. The Ultrasonic Distance Sensors were mounted on the electronics plates and the limit switches were directly attached to the 80/20 aluminum extrusion that made up the robot's frame. During our final evaluation, our limit switches had a little difficulty interfacing with their intended contact points. Future work could include a custom limit switch housing that incorporates a wider contact point that facilitates the activation of the switch when pressed against a flat surface like those on the stairs.