Autonomy work for outside teams.
AV@B Applied takes on autonomy, robotics and AI projects for startups, labs and engineering teams.
You get Berkeley engineers who build autonomy every week on our own vehicles. We get real problems to ship, and the fees pay for the rest of the club.
What we take on
The same skills our vehicles depend on, applied to your robot, product or research.
Perception
Detection, segmentation, depth and sensor fusion across cameras, lidar and radar.
Robotics and ROS
ROS and ROS 2 systems, autonomy software and the engineering that keeps a robot reliable.
Simulation
Simulation infrastructure for testing behavior before it reaches hardware.
Fast inference
Model deployment, edge AI and GPU inference optimization.
Data and evaluation
Dataset and evaluation pipelines that tell you whether a model got better.
Robotics engineering
Integration work across hardware, sensors and software.
Problems like these
Examples of the kind of work we can take on. If yours looks different, ask anyway.
Our robot loses track of where it is indoors.
Localization ›Sensor fusion and localization that hold up without GPS.
Our detector is too slow on the edge device.
Inference ›Profiling, quantization and GPU optimization to hit your latency budget.
We can't tell if the new model is better.
Evaluation ›Datasets, metrics and pipelines that make model changes measurable.
We need to test before touching hardware.
Simulation ›A simulation setup that runs your stack against realistic scenarios.
Our ROS system is held together with tape.
ROS ›Cleaner architecture, launch setup and tooling for a ROS 2 stack.
How a project works
From first call to handoff.
- 01
Tell us the problem
- 02
Scope it together
- 03
We staff a team
- 04
Build and review
- 05
Hand off the work
The money goes back into building.
- ComputeGPUs for training and simulation
- SensorsLidar, cameras and radar
- VehiclesNew platforms as we move up in size
- Bigger projectsWork the club could not fund otherwise
Tell us what you're stuck on.
We work on autonomy, robotics, computer vision, ML infrastructure and fast inference.
Describe what you're building and where it's stuck.