Platform guide
Everything a university, research laboratory or competition team needs to know about the RoboRacer Core Kit — the 1:10-scale autonomous racing robotics kit that arrives assembled, wired and running ROS 2 on an NVIDIA Jetson Orin Nano.
Definition
An autonomous racing robotics kit is a small-scale vehicle carrying the complete sense–plan–act loop of a full-size self-driving car. A scanning sensor measures the track, an onboard computer localises the vehicle and plans a trajectory, a motor controller executes steering and throttle commands, and a power distribution board keeps all three supplied from a single battery. Because the vehicle is 1:10 scale, a team can run genuinely aggressive control experiments — at the limit of tyre grip, at ten metres per second — in a corridor or a sports hall rather than on a proving ground.
The RoboRacer Core Kit from Ambimat Electronics is that vehicle, delivered finished. It
is built to the 1:10-scale specification used by the RoboRacer community — the international autonomous
racing programme widely known by its earlier name, F1TENTH, whose open-source code still
lives under the f1tenth GitHub organisation and whose Autonomous Grand Prix events run
alongside robotics conferences such as IEEE ICRA and IROS. Choosing that specification matters more than
it first appears: it is what makes published research, competition code and teaching material from other
institutions run on your vehicle without a hardware redesign.
Audience
Three groups buy this kit, and they buy it for the same underlying reason: the interesting work is in the algorithms, and hardware integration is the tax standing in front of it.
A robotics, mechatronics or control course needs every team on the same platform, or laboratory sessions become hardware debugging sessions. Identical, pre-validated vehicles mean a lecturer writes one set of notes, one marking scheme and one reproducible baseline that behaves the same in week one and week twelve.
Teams preparing for a RoboRacer / F1TENTH Autonomous Grand Prix are judged on lap time and overtaking, not on soldering. Starting from a vehicle that already localises and drives means the season is spent tuning pure pursuit, disparity extension, Frenet-frame planning or a reinforcement-learning policy against the clock.
Reviewers increasingly expect a physical demonstration alongside simulation results. A standard 1:10 platform gives a laboratory a hardware baseline other groups can reproduce, and enough compute headroom on the Jetson to run learned perception, model-predictive control or multi-agent experiments onboard.
Bill of materials
The kit ships as one assembled vehicle. Every item below arrives mounted, wired and tested together rather than as loose parts to integrate.
| Component | What it does on the vehicle |
|---|---|
| NVIDIA Jetson Orin Nano Super | Onboard compute. Runs Ubuntu, the ROS 2 graph and any GPU-accelerated perception or learned policy. |
| Hokuyo 2D LiDAR | Primary perception. 270° planar scan over Ethernet, feeding SLAM, localisation and obstacle avoidance. |
| TRAMPA VESC 6 MkVI | Motor controller. Field-oriented drive of the brushless motor, steering servo control, wheel odometry and the onboard IMU. |
| Power distribution unit | The Ambimat AE170 PDU. Converts battery or wall-adapter input into the regulated 19 V, 12 V and 5 V rails the rest of the vehicle runs on. |
| Traxxas Slash 4X4 Brushless chassis (P/N 68454-4) | The 1:10-scale rolling platform — shaft-driven four-wheel drive and the Velineon V3200 brushless power system (3200 Kv). Manufacturer specifications. |
| 3S lithium-polymer battery (India only – not for export) | Vehicle power for both traction and, through the PDU, the electronics. |
| Battery charger (India only – not for export) | Balance charger for the 3S pack. |
Technical summary
| Vehicle class | 1:10-scale autonomous racing vehicle, RoboRacer / F1TENTH specification |
|---|---|
| Compute | NVIDIA Jetson Orin Nano Super — 1024-core Ampere GPU with 32 tensor cores, 6-core Arm Cortex-A78AE CPU, 8 GB LPDDR5 |
| Perception | Hokuyo 2D laser scanner — 270° field of view, 40 Hz scan rate, Ethernet interface |
| Actuation | VESC 6 MkVI field-oriented motor controller with integrated IMU; Traxxas Velineon V3200 brushless drive (3200 Kv) |
| Power rails | Ambimat AE170 PDU V02.04 — 19 V at 3 A, 12 V at 2 A, 5 V at 1 A |
| Operating system | Ubuntu 22.04 LTS with NVIDIA JetPack, pre-flashed. Included in the INR 628,000 + GST option. |
| Middleware | ROS 2 Humble Hawksbill, configured and validated before dispatch. Included in the INR 628,000 + GST option. |
| Software Setup / Technical Support | Operating-system loading and ROS stack configuration, plus 60 days of remote hardware and software technical support after delivery. Included in the INR 628,000 + GST option. |
Software
Nothing in the stack is proprietary to Ambimat Electronics. The value is that it is already installed, already talking to the hardware and already validated on the exact vehicle you receive.
Ubuntu 22.04 LTS with NVIDIA JetPack, which brings the Linux for Tegra
kernel, CUDA, cuDNN and TensorRT. That pairing is deliberate: 22.04 is the Tier 1 platform for
ROS 2 Humble Hawksbill, so binary ROS packages install with apt rather than
being built from source.
Where Software Setup / Technical Support is included, the vehicle boots with its ROS 2 graph configured —
the LiDAR driver publishing
sensor_msgs/LaserScan, the VESC driver publishing odometry and accepting drive commands, and
a transform tree describing the vehicle's frames. Work proceeds through the ordinary ROS 2 toolchain:
rclpy or rclcpp for nodes, colcon for builds,
ros2 bag for recording experiments, and RViz2 for visualising scans, maps
and planned paths. DDS handles transport, so a laptop on the same network joins the graph as a monitoring
station.
What runs on top is your research. The platform is the one the RoboRacer / F1TENTH community targets, so the standard progression of algorithms is directly available:
slam_toolbox or
Cartographer to build an occupancy grid, then particle-filter localisation against that map at race
speed.
f1tenth_gym simulator and transferred to the vehicle, with inference accelerated on the
Jetson's GPU through PyTorch or TensorRT.
Decision
Sourcing the same bill of materials yourself is entirely possible — the RoboRacer / F1TENTH build documentation is public. The honest comparison is not about capability, it is about where a team's first two months go.
| Consideration | Building from parts | RoboRacer Core Kit |
|---|---|---|
| Procurement | Eight or more suppliers across several countries, each with its own lead time, import duty and minimum order. A single back-ordered part stalls the build. | One purchase order, one supplier, one delivery. |
| Power electronics | Design or adapt a distribution board that feeds 19 V, 12 V and 5 V from one pack without browning out the Jetson under motor current draw. A common source of intermittent, hard-to-diagnose faults. | The AE170 PDU, already specified, built and tested against this exact load. |
| Assembly and wiring | Mechanical fit-out, connector crimping, strain relief and cable routing that survives a crash. | Done, on a laser-cut MDF base plate. |
| Software bring-up | Flash JetPack, resolve driver and kernel version conflicts, calibrate the VESC, configure the LiDAR network interface, build the ROS 2 workspace. | Hardware bring-up is done. With Software Setup / Technical Support the Jetson is also flashed and the ROS graph publishes on first boot; without it, the software stack is yours to build. |
| Time to first autonomous lap | Typically several weeks of elapsed calendar time, concentrated in whoever knows the hardware. | The same afternoon the crate is opened. |
| Fleet consistency | Vehicles drift apart as parts are substituted between orders, so results stop being comparable across teams. | Identical builds, so a result on one vehicle means something on another. |
| When it is the wrong choice | If the integration is the syllabus — an electronics or systems-engineering course where students should design the harness and the power tree themselves. | If the autonomy is the syllabus, or the deadline is a competition date rather than an end of term. |
Questions
An autonomous racing robotics kit is a small-scale vehicle that carries everything needed to drive itself: a sensor for perception, an onboard computer for planning and control, a motor controller for actuation, regulated power for all three, and a software stack that ties them together. The RoboRacer Core Kit is a 1:10-scale example built to the RoboRacer / F1TENTH specification, using a Hokuyo 2D LiDAR, an NVIDIA Jetson Orin Nano, a TRAMPA VESC 6 MkVI motor controller and ROS 2 on a Traxxas Slash 4X4 Brushless (P/N 68454-4) chassis.
Yes. The kit is built to the 1:10-scale RoboRacer / F1TENTH vehicle specification and uses the component
classes that community's open-source stack expects — a Traxxas Slash 4X4 Brushless (P/N 68454-4)
chassis, a VESC motor controller, a 2D scanning LiDAR and an NVIDIA Jetson compute module. Code
published under the
f1tenth GitHub organisation, including the f1tenth_gym simulator, runs against
it without a hardware redesign.
The standard build pairs an NVIDIA Jetson Orin Nano Super — a 1024-core Ampere GPU with 32 tensor cores, a 6-core Arm Cortex-A78AE CPU and 8 GB of LPDDR5 — with a Hokuyo 2D laser scanner giving a 270° field of view at 40 Hz, and the inertial measurement unit integrated into the VESC 6 MkVI motor controller. Exact sensor and compute variants are confirmed on your quotation.
Yes. Where Software Setup / Technical Support is included, the vehicle is delivered with the Jetson
operating system already flashed — Ubuntu 22.04 LTS with NVIDIA JetPack — and a configured ROS 2 stack,
so the LiDAR, VESC and vehicle transform tree are publishing on first boot. Teams work in
rclpy or rclcpp, visualise in RViz2, record runs with
ros2 bag and build with colcon exactly as they would on any other ROS 2 robot.
No. Component selection, procurement, mechanical assembly, wiring, power distribution and software validation are all completed by Ambimat Electronics before dispatch. The vehicle arrives built and tested; the remaining work is charging the battery, joining the vehicle to your network and starting the stack.
Yes. The kit is a platform, not a closed product. The ROS 2 stack is ordinary open-source software you can replace node by node, and the Ambimat AE170 power distribution board exposes spare regulated 12 V and 5 V outputs for additional hardware such as a depth camera or a separate IMU. The board's design files are published as open source.
It covers loading of the operating system for the NVIDIA Jetson, complete ROS stack configuration including the LiDAR and VESC drivers, validated on the vehicle before dispatch, plus remote hardware and software technical support for up to 60 days post delivery to assist with deployment and setup.
Software Setup / Technical Support is included in the INR 628,000 + GST option. INR 580,000 + GST is the same complete physical kit supplied for customer-managed software setup and integration: the Jetson arrives without the operating system flashed, the ROS stack and the LiDAR and VESC drivers are not configured, and the 60-day support period does not apply.
The US$5,870 international price is on the same customer-managed basis. International customers who require Software Setup / Technical Support can purchase the RoboRacer Core Kit Pro (with Software Setup / Technical Support) for US$6,250.