Depth Vision Powered by Orbbec Gemini 336L

The Gemini 336L stereo vision camera forms the core of the depth perception system in every Model C2 robot, enabling reliable navigation through dynamic and unstructured environments. This compact sensor provides high-resolution depth and RGB data that supports real-time mapping, obstacle avoidance, and zone recognition critical for cleanroom compliance.

Rather than focusing on the hardware alone, Quasi Robotics emphasizes the integration of Gemini 336L into its full-stack autonomy system. For technical specifications and performance characteristics of the camera, click here to learn more about Gemini 336L.

Integration: Design for Reliability and Speed

Mechanical & Electrical Integration

The Gemini 336L was embedded directly into Model C2’s sensing unit using a standard 1/4-20 mount. Its lightweight form factor (135g) and IP65-rated enclosure enabled integration without reconfiguring the robot’s body or protective constraints. The USB 3.0 interface offered plug-and-play connectivity to C2’s onboard compute system.

Software Integration

Using the Orbbec SDK, Quasi engineers integrated real-time synchronized RGB and depth streams into their Q.AI perception stack, which runs alongside ROS-based modules. The camera’s stereo depth stream feeds into algorithms for:

• 3D SLAM and real-time localization
• Dynamic object tracking and predictive path planning
• Identification of sterile vs. non-sterile zones for cleanroom compliance

Application Scenarios

Model C2 is tailored for environments where precision, sterility, and adaptability are non-negotiable:

• Life Sciences & Biopharma: Autonomous handling of reagents and sensitive materials
• Healthcare: Safe and contactless delivery of meals or supplies in clinical settings
• Manufacturing: Intra-facility transport of tools and components
• Retail & Hospitality: Automated restocking and guest services in high-traffic areas

Integrated Benefits of Gemini 336L in Model C2

a. Depth-Enabled Localization and Navigation in Tight Spaces
The Gemini 336L provides dense 3D depth data that supports high-precision localization and agile maneuvering in constrained environments. This allows C2 to operate confidently in narrow corridors, around corners, and within dense lab or factory layouts where centimeter-level perception is essential.

b. Stable Object Tracking for Dynamic Environments
By leveraging stereo depth information, the camera enhances real-time tracking of moving objects and people. This stability helps the robot anticipate trajectories, maintain safe distances, and make smooth, proactive adjustments—especially in shared workspaces with frequent human activity.

c. Extended Field-of-View for Early Hazard Detection
The stereo configuration of the Gemini 336L provides an expanded perspective, enabling C2 to identify potential obstacles before they enter the robot’s direct path. This is particularly beneficial when navigating around blind corners, through doorways, or past equipment in dynamic environments.

d. Lighting and Environment Adaptability
Gemini 336L’s ability to maintain accuracy in sunlight and low-light conditions enables 24/7 operation across indoor and semi-outdoor areas.

e. Plug-and-Play Scalability
With USB UVC compatibility and a robust SDK, Gemini 336L supports rapid deployment without adding complexity. This complements C2’s “no-integration-needed” promise, making it easy for users to scale across multiple sites or applications.

The collaboration between Quasi Robotics and Orbbec showcases how well-integrated 3D vision can elevate AMR capabilities in real-world, high-stakes environments. By embedding the Gemini 336L into its Model C2 platform, Quasi has built a system that’s fast to deploy, safe to operate, and scalable across industries.

“The Gemini 336L was a natural fit for Model C2. Its plug-and-play design, depth fidelity, and robust SDK gave us the foundation we needed to scale our Q.AI perception stack without compromising C2’s deployment promise. It’s enabled us to move faster, smarter, and safer—exactly what our customers expect.”

——Quasi Robotics Team