Real-time policy execution on Jetson-class compute
LimbBot Core supports this capability as part of the runtime layer for whole-body control programs.
On-robot whole-body control runtime for balance, locomotion, and payload-aware motion at the edge.
Runs the approved control policy on robot compute and keeps balance decisions close to the hardware.
Capabilities are scoped for robotics teams building pilots and production fleets.
LimbBot Core supports this capability as part of the runtime layer for whole-body control programs.
LimbBot Core supports this capability as part of the runtime layer for whole-body control programs.
LimbBot Core supports this capability as part of the runtime layer for whole-body control programs.
LimbBot Core supports this capability as part of the runtime layer for whole-body control programs.
Provide robot model, interfaces, and deployment constraints.
Set product-specific parameters, permissions, and success metrics.
Run simulation, lab, or telemetry checks before promotion.
Use logs and reviews to improve the next policy or release.
| Area | Details | Notes |
|---|---|---|
| Primary role | Runtime | On-robot whole-body control runtime for balance, locomotion, and payload-aware motion at the edge. |
| Runtime path | Edge or cloud depending on product | real-time control stays on robot for Core |
| Interfaces | ROS 2, SDK, REST, telemetry | varies by program |
| NVIDIA stack | Isaac, Jetson, CUDA/TensorRT, OpenUSD | used where relevant |
Visual placeholder showing how this product fits in the LimbBot control platform.
Use cases depend on program maturity and robot body class.
Validate walking and recovery over industrial routes.
Support balance during payload and tool tasks.
Benchmark morphologies and control approaches.
Package reusable control into shipped robot programs.
Pricing depends on body class, product mix, fleet size, and cloud usage. Most teams start with a pilot scope.
LimbBot Core is for robotics teams that need the runtime capability described on this page.
It can be evaluated alone, but production programs often combine multiple LimbBot products.
Robot model, compute target, interfaces, and a bounded first task.
Policies and configurations move through validation and safety review before robot rollout.
Most deployments combine runtime, safety, simulation, cloud adaptation, planning, and API integration.
Cloud fine-tuning and dynamics adaptation for new bodies, payloads, terrains, and tasks.
View productGPU simulation and reinforcement-learning pipeline built on NVIDIA Isaac for contact-rich robot motion.
View productSafety layer for fall avoidance, force limits, stop modes, and event telemetry around learned control.
View productAgentic loco-manipulation motion planner that composes sub-skills for long-horizon mobile manipulation.
View productDeveloper SDK and integration toolkit with ROS 2 node, REST controls, telemetry, and deployment utilities.
View productShare your robot class and first task to scope integration.
Request access