THE MISSION
Delivery is building autonomous robots that bring food, drinks, and packages to students and faculty across campus. Started in August 2025, the project focuses on real impact on campus life. Our robots use GPS, inertial, and wheel sensors to navigate campus paths and sidewalks.
Software backbone
Precision positioning
Car-style steering
IN ACTION
Watch the robot out on campus, then see the autonomy stack drive itself through a digital twin of east campus.
On campus
Rolling across the quad through a crowd of students.
In simulation
The full autonomy stack driving itself through east campus in Gazebo: localization, planning, and control all running together.
THE CODE
Everything we build is open source. Dig into the code, fork it, and come build it with us.
LIVE
BearCarts
Where campus orders start.
bearcarts.com
TEAM REPO
delivery-robo
Our monorepo: ROS 2 workspace, Arduino firmware, Gazebo simulation, and deployment.
WUSTL-Delivery/delivery-robo
AUTONOMY STACK
delivery-autonomy
EKF sensor fusion, OpenStreetMap sidewalk routing, and MPPI control, all in simulation.
Built by Sebastian (@sebtheiler)
sebtheiler/delivery-autonomy
Three subteams, one robot. Here’s how software, electrical, and mechanical come together.
Sensor fusion
An Extended Kalman Filter blends GPS, IMU, and wheel-encoder data into one confident position estimate.
Sidewalk routing
A global planner builds routes straight from OpenStreetMap sidewalk data for east campus.
MPPI control
A Model Predictive Path Integral controller steers the Ackermann drivetrain smoothly along the route.
Digital twin
A Gazebo simulation with a full ros2_control Ackermann model, so every change gets tested before it hits pavement.
Robot platform
ROS 2 nodes for the IMU, RTK GNSS corrections, and joystick teleop, plus a command mux that always lets a human take over.
Up next
SLAM and Nav2, lidar and depth-camera obstacle avoidance, and cost maps that keep the robot on the path instead of the grass.
Compute
A Raspberry Pi runs ROS 2 and talks to the motor controller over a lightweight serial command protocol.
Motor control
Custom Arduino firmware drives the motor, positions the steering servo, and reads the wheel encoder.
Sensing
A BNO08x IMU over I²C plus a u-blox RTK GNSS receiver for high-precision positioning.
Power
Onboard battery monitoring with a live voltage readout on an LCD.
Up next
An RPLIDAR 2D lidar, an Intel RealSense depth camera, and a magnetic encoder for closed-loop steering.
Ackermann chassis
Four wheels with steerable front knuckles, the same geometry as a car, for smooth turns on campus paths.
Drivetrain
A single drive motor for propulsion and a servo that handles the steering.
6-inch wheels
Sized to roll over sidewalk seams, curb cuts, and the occasional patch of grass.
Cargo bay
A tote-style bin rides up top to carry the goods.
RIGHT NOW
Three goals we’re chasing before the semester wraps up.
01
A new diff-drive robot
Design and build a new differential-drive robot that can turn in place and handle tight campus paths.
02
Autonomy on the real robot
Take the autonomy stack out of simulation and onto hardware: localization, planning, and control running on real campus paths.
03
A real delivery
Complete a real, end-to-end delivery on campus this semester.
THE BIG PICTURE
01
Live on campus
Test through the fall, then run a live delivery service during campus events like WILD.
02
Full autonomy
Move the simulated stack onto the real robot and drive fully autonomously on campus paths.
03
Order & track
Ship the ordering app and backend so students and faculty can request a delivery and follow it.
04
A whole fleet
Scale from one robot to a fleet handling many deliveries across campus at once.
OUR STACK
ROS 2
Python
C++
Gazebo
ros2_control
Arduino
Raspberry Pi
RTK GNSS
BNO08x IMU
Extended Kalman Filter
MPPI control
OpenStreetMap
Nix
Git & GitHub
JOIN THE TEAM
We welcome anyone who wants to build something people on campus will actually use. No one needs to know everything. Relevant skills include:
Software
Software development (Python/C++), ROS 2, and the terminal
Robotics
Robotics fundamentals: localization, planning, and control
Backend & systems
Backend and systems design: pub/sub, APIs, and app development
Mechanical
Mechanical design, CAD, and 3D printing
Electrical
Electrical engineering, embedded firmware, and PCB design
Mindset
Curiosity, rapid learning, and working independently and collaboratively