RVPU perception for ground & mobile platforms
Configurable vision for robots that work where the network does not.
Dense stereo 3D from the cameras you chose, automatic self-calibration, object detection and 3D tracking, and thermal–RGB matching — perception functions you configure and run, on an AMD Kria KV260. Wire your rig in, get 3D back. No tethered GPU. No cloud round-trip. No EdgeAI team.
Use cases
Where on-platform vision earns its keep
Dense 3D from your own rig
Stereo and multi-view depth computed from the cameras, baseline, and field of view you picked — not from a locked vendor stereo SKU that fixes those decisions for you.
Automatic self-calibration
After a first calibration against reference frames, the unit re-calibrates itself to compensate for drift — so a rig that shifts under vibration keeps producing usable depth.
Detection, segmentation & 3D tracking
Detect and segment objects, characterize selected classes, place them in a 3D map, and track them over time — with the results feeding your planner directly over ROS 2.
Cross-modality matching
Match thermal against RGB and run follow-on inference only on the matched regions — one scene representation out of several sensors, computed on the unit.
Comms-denied autonomy
Perception stays on the robot when the link drops. No degradation when the uplink does — resilient in contested environments.
Sovereign & resilient
Sovereign supply and defense-grade posture — deployable under any country’s sovereignty requirements, with hardware root of trust, signed firmware, per-module attestation, and a forensic trace buffer. Designed for MIL-STD-810H environments — qualification on the roadmap.
The matched RVPU
TerraBot X
RVPU · ground robotics · Kria KV260
TerraBot X is an RVPU: a set of perception functions you wire cameras into and get useful outputs from. Choose your own cameras, baseline, and field of view, connect them through the hardware interface, and get dense 3D, tracked objects, and cross-modality matches back. No vendor-locked stereo kit, no compiler toolchain to learn, no EdgeAI team to hire.
- Platform
- AMD Kria KV260
- Perception
- Dense 3D · detect · track
- Cameras
- Yours — any geometry
- Operating range
- −40 to +85 °C target
Why this module, this vertical
The properties that map to your platform.
Perception as a unit
The RVPU is not a generic AI accelerator. It is a curated set of robotics perception functions — dense 3D, detection, segmentation, tracking, cross-modality matching — exposed as configured pipelines. You wire cameras in and get useful outputs back, without building a perception stack first.
Your cameras, your geometry
Off-the-shelf 3D platforms force a vendor camera kit and fix your baseline, placement, and field of view. The RVPU does the opposite: pick the cameras your SWaP, cost, and supply chain allow, place them how your platform needs, and the unit works with that rig.
Calibrates itself
After one calibration against reference frames, the unit re-calibrates automatically to compensate for drift. A rig that shifts under vibration or thermal cycling keeps producing depth you can trust — no field re-calibration procedure.
Secure by design, forensic by default
Hardware root of trust, signed firmware, and per-module attestation keep the device tamper-evident. An 8,192-entry cycle-accurate hardware trace buffer gives forensic-grade observability for mission review and ROE compliance.
Invotet SDK
Describe your rig. Run the pipeline.
A Python SDK for configuring the RVPU — describe your cameras and their geometry, pick the perception functions you need, and stream 3D positions, depth maps, and tracks into ROS 2. It ingests PyTorch, ONNX, and HuggingFace models when a pipeline needs a custom detector, with no CUDA in the loop. App config, not model code.
Framework
PyTorch
Bring a custom detector: trace or torch.export models fold into a pipeline with no rewrite.
Framework
ONNX
Standards-based interchange — any ONNX-exported model can back a pipeline stage.
Framework
HuggingFace
Vision checkpoints load through a one-line loader when a pipeline is customised.
Talk to the team behind the RVPU.
Most robotics conversations start with your cameras and your autonomy stack. Tell us the rig — how many cameras, what baseline, which modalities — and what your robot has to see, and we will set up a TerraBot X evaluation through the design-partner program and route the right documentation.