January - June 2020
Concept creation | Virtual prototyping | Dynamic Simulation | Manufacturable part design | Rendering
Solidworks | Blender
This project was an exercise of CAD and technical development. Inspired by the original MiniWheg robot from the Biologically Inspired Robotics Lab at Case Western Reserve University. The specification required the robot to have four "wheel legs" capable of climbing steps up to 55 mm tall, traverse uneven terrain, travel at least 0.5 m/s and have a turning circle less than 160 mm. The robot could not weigh more than 200 g and should survive a fall from 300 mm. The project took the form of a 20 page report with prototyping, testing, development and simulation.
Key stages of research development and testing
I began the project looking for inspiration for a design and approach. Continuing with bio-mimicry, I was interested in making an insect inspired robot with a focus on agility and an extremely compact body.
My project centred around the design of the wheg (wheel - leg). Once the wheg was designed a chassis and system were created to drive them. I wanted the wheg to be very circular so that the robot could drive with minimal up and down wobbling. There also had to be sufficient 'hook' spaces to latch onto obstacles and steps and pull the robot up. After initial rounds of virtual prototyping I chose a three legged spiral design with small sharp hooks, inspired by insect legs which help grip uneven surfaces while climbing.
Towards the end of the project, the whegs were modified to improve strength and reduce weight. The image shows the light-weighted design in a drop test with a 400 g to simulate the robot being dropped on a single wheg from 350 mm.
I used the prototyping stage to identify key physical challenges. I looked into the step climb, turning circle, and component layout in the chassis. I found shorter chassis and closer whegs helped climb steps. The turning circle presented challenges and would need a sharp turning angle to achieve the turning circle. Instead, to optimise agility, I chose to drive the left and right whegs independently, which would also allow for on the spot turning. This meant two motors and two gear drives were needed, making component placement more challenging. I gained a rough idea for a layout and moved to component selection.
With an idea for the size, weight and dual motor function, I researched different ways to achieve the outcome. Calculation helped specify power requirements, for example considering the torque required to lift the robot up a step, or the rotational speed and wheg diameter to achieve 0.5 m/s. Several iterations of component choices were made, honing in on a balance of power and weight. Different motors, gear ratios and power sources were calculated and considered. The chosen system used an ultra slim "sports" Li-Po battery with high current output. Dual compound gears step up the torque to required levels while ensuring speed was acceptable. Power from the driven wheg was transferred to the passive by timing belts and pulleys for a light weight, high torque system.
CAD: Chassis design
I began by creating the components as CAD models, allowing me to experiment with arrangements. With the target to be extremely compact, I arranged the components into as small a space as possible, and built the casing around. The casing just large enough, with the gears fitting into the rounded front and back. Starting from the centre, mouldings are made to hold the components in place. This meant when all components were placed side to side, any excess length of chassis could be removed.
The drive system rotates on steel shafts which sit on mouldings in the case, acting as journal bearings. This was suitable for the miniature size of low mass of the device. All shafts rotate on the split line of the casing. This makes it easy to have semi circle mouldings which line up to form complete bearings.
The chassis was designed to be assembled from one orientation. As the exploded view shows, parts are fitted sequentially from the top, followed by the lid.
Testing: Obstacle course and failed step climb
Throughout the process, extensive dynamic tests were used to identify weaknesses so they could be resolved. Testing was also done to validate performance virtually before any production took place (this was also an entirely virtual project).
I made a short obstacle course which combined various steps, climbs, and falls to test if the combination of chassis length, depth, and wheg shape could traverse the course. The first iterations of the robot could not climb the step, and would get stuck part way up. This confirmed that a shorter chassis was needed, and would eventually lead to a design where the front and rear whegs overlapped each other.
Testing: Obstacle course completion
Modifying the internals to shorten the chassis and enabling faster wheg rotations allowed the robot to complete the course. I found that this iteration of the design could climb the step if the motors ran over 175 RPM. At this speed, the whegs made enough impact on the ground to kick the robot up just enough for the rear whegs to grab the step
Testing: Final redesign
Going into more testing depth, I realised that while the motors could run at the 200 RPM, upon climbing the step the torque requirement increases dramatically beyond the motor ability. The next redesign therefore was concerned with getting over the step without exceeding torque limits. By shortening the chassis further such that the whegs overlapped, the robot could climb the step without relying on "jumping".
Testing: Rough terrain
A quick test over a generated rough terrain helps confirm the robot's ability to drive smoothly and turn on the spot
A final plastics study is used to check if the designed chassis parts are able to be injection moulded. The study simulates the injection of ABS, and checks for fill time, pressure, front flow temperature, and other measures which determine if a part would fail in manufacture. This stage revealed a few small things to modify, such as reducing some wall thicknesses.