01
01 Background
The Major Qualifying Project, or MQP, is the capstone design project at WPI that challenges students to apply their technical knowledge to a substantial engineering problem. It represents the culmination of an undergraduate engineering education.
Our team retrofitted a 1995 Club Car golf cart with autonomous vehicle systems through modular construction and engineering. We improved upon the existing steering system, ruggedized the braking system, and automated the accelerator input. A graduate student partnered with our team and provided software for stereoscopic vision systems, image processing, and mobile path planning. This platform is now prepared for future teams to develop software, mobile applications, and use the autonomous vehicle for various user applications.
02
02 Student Researchers
Robert Crimmins — WPI senior, double major in Electrical and Computer Engineering and Robotics Engineering. Mechanical focus: braking, steering, power systems, and chassis.
Raymond Wang — WPI senior, Robotics Engineering. Software focus: throttle control and stereoscopic vision.
Guilherme Meira — WPI Masters student, Wireless Innovation Lab. Stereoscopic vision, image processing, and path planning.
03
03 Project Advisor
Professor Alexander Wyglinski — Electrical and Computer Engineering / Robotics Engineering, Wireless Innovation Laboratory at Worcester Polytechnic Institute.
04
04 System Architecture
Robocart requires many subsystems working simultaneously to reach safe and reliable autonomy. Our group broke down the project into seven subsystems: Throttle, Brake-by-wire, Steer-by-wire, Chassis Reinforcement, Power Systems, Server, and Stereoscopic Vision. Each subsystem works independently as its own system, yet comes together to make the entire vehicle functional.
By tackling the mechanical, electrical, and software components, as well as reinforcing the chassis and making improvements to the server and power systems, we brought the golf cart together as an untethered vehicle capable of autonomous operation.
05
05 Throttle-by-Wire
The original throttle system operated on a continuously variable potentiometer that changed its value from 5500 Ω when the acceleration pedal is at rest, to 0 Ω when the pedal is fully compressed. To electronically replicate this, we implemented an Addressable Dual Digital Potentiometer from Maxim, specifically the DS1803-010, communicating with an Arduino microcontroller over I²C protocol.
Since the DS1803 uses an 8-bit value (256 steps) to control 10,000 Ω of potentiation, we achieve control within ~39 Ω accuracy per increment. With the golf cart’s maximum velocity of 20 mph, this gives us precise speed control for both acceleration and deceleration. A quad-pull-quad-throw switch allows manual switching between autonomous and manual driving modes.
06
06 Brake-by-Wire
The golf cart uses a drum brake mechanism to slow the wheels. The manufacturer fitted a brake wire system applied by pressing the brake pedal, putting tension on the wire, and extending the brake shoes against the drum. This system is purely mechanical with no means of autonomous operation.
To prepare for autonomy, our team designed and created a custom brake coupler machined on a lathe from ¾″ aluminum stock, attached to a solid steel base housing a metal bearing for smooth rotation. A windshield wiper motor drives the coupler, which tensions vinyl-coated steel braided cable connected to the existing brake lines. The chassis I-beam was modified with an angle grinder and reinforced with fitted steel plates to accommodate the assembly.
In our final design, we drilled through the aluminum coupler and the motor axle, securing them with an M3 bolt and Nylock washer to eliminate any slipping during repeated use. The motor controller, a Dimension Engineering Sabertooth 2x60, supplies power to both the steering and braking motors via microcontroller commands.
07
07 Steer-by-Wire
The steer-by-wire system from the previous academic year had alignment issues with the steering column and acrylic plates. Our team replaced the acrylic with a new aluminum backplate, cutting a 4″ diameter hole with a hole saw to accommodate the steering column bearing. We machined an Acetal Delrin plastic coupler for the steering assembly and 3D-printed a potentiometer bracket in AutoCAD for position feedback.
The steering system uses a chain and sprocket mechanism driven by a pancake motor. A 20-turn potentiometer provides precise angular feedback to the microcontroller, enabling accurate autonomous steering control. Safety considerations include preventing over-rotation of the rack and pinion system, which could cause separation and make steering impossible.
08
08 Additional Enhancements
Body Restoration: We sanded, primed, and painted the body panels at a professional auto body shop using DuPont paint mixed to match WPI’s official Pantone 187c red, finished with gloss, pearls, and a protective clear coat.
Front Rack & Camera Mounts: We 3D-printed brackets with hex nut traps to mount 80/20 T-slotted aluminum extrusion as a modular front rack. Raspberry Pi computers and camera modules were mounted on custom 3D-printed brackets, allowing easy adjustment and removal for servicing.
Server: To eliminate wireless latency issues faced by previous teams, we mounted the server directly on the golf cart. A cargo utility rack was built from 80/20 aluminum with 3D-printed pivot brackets—an 11-fold cost savings over commercial equivalents. The server ran Ubuntu 14.04 with all microcontrollers communicating over USB and Raspberry Pi cameras over an Ethernet switch.
Power Systems: We upgraded to deep-cycle 29HM batteries selected through a performance matrix evaluating capacity, charge density, volume, and price. Custom wood floor pans were cut, painted, and bolted to the chassis to support the batteries. All electrical systems were consolidated to a 24-pin ATX connector for easy diagnostics, calibration, and transport.