中文
A home embodied robot interacting with a family

Home Embodied Robot Solutions

DEPTRUM pairs the Nebula 411 with HyperCube 3D SLAM for indoor mapping, continuous localization and navigation in home robots. Front and rear cameras extend coverage when the robot turns or reverses, even in compact bodies with limited compute.

Contact Us

The Challenge

A home is full of complex furniture and obstacles at different heights. The robot has to understand the 3D environment and keep confirming its own position as it moves between rooms. Blind spots when turning or reversing, plus a tight body and limited compute, make autonomous navigation even harder.

3D spatial perception in a home platform environment

Complex spatial structure

Furniture and obstacles vary in height, so 2D outlines fall short

Cross-room localization

Continuous movement needs mapping and position estimation to run together

Blind spots when turning and reversing

A forward view cannot cover turns and reversing, so extra observation is needed

Limited space and compute

A compact body and limited compute demand a balance of performance and cost

Solution Architecture

System ArchitectureUser ScenarioHardwareDepth SDKApplication SDKHome Embodied RobotIndoor mobile platformApplication NeedsAvoidance · navigationNebula 411 · FrontForward-facingNebula 411 · RearRear-facingmToF Depth ProcessingmToF point cloudMultipath resistant · low motion blurHypercube SLAM3D mapping and positioningFront/rear point clouds · IMUTriggerFunction DeliveryData Flow* HyperCube: DEPTRUM point-cloud 3D SLAM

Recommended Products

Ask about these products

Solution Advantages

Depth perception in a home environment

Compact integration

Miniature modules fit into a robot's exterior and internal structure and can be mounted front and rear, reducing the constraints sensing hardware places on industrial design.

Near and far sensing together

sToF and mToF are combined to cover both close obstacles and room-scale surroundings, providing the perception needed for 3D mapping, localization and autonomous movement.

Front and rear views complement each other

Front and rear cameras add observations of the environment, reducing blind spots when turning or reversing and giving more complete spatial data for movement in any direction.

Simpler hardware

Integrated depth perception supports mapping and obstacle avoidance, cutting sensor count and optimizing structure, wiring and hardware cost.

Low power, lightweight compute

Proprietary components and lightweight depth algorithms reduce power draw and compute load, leaving resources for interaction, task models and control.

Application Cases

Noetix Robotics home embodied robot case study

Noetix Robotics

Powered by Deptrum's embodied 3D SLAM solution, delivering spatial mapping, localization & navigation, and intelligent obstacle avoidance

FAQ

How many cameras are needed for mapping and navigation?
The number depends on the robot's field of view, how it moves and the scope of its tasks. A single camera can support an appropriate mapping and avoidance setup, while front and rear cameras add observations when turning or reversing; the final configuration depends on the robot's structure and compute resources. We recommend one Nebula camera at the front and one at the rear with hardware exposure synchronization for better environmental adaptability.
How do multiple cameras work together?
Synchronization and timing settings organize multi-camera capture. Camera count, field-of-view overlap and mounting position all affect system performance, so optical interference, data synchronization, communication bandwidth and algorithm behaviour should be validated against the actual layout.
Does the solution support ROS2?
ROS2 integration support is available per project. Integration requires confirming the target ROS2 distribution, host platform, operating system and camera interface, and matching the corresponding drivers, data formats and software versions.

Building the Visual Perception Infrastructure for the Physical AI Era

Contact Us