Opteran turns biological brain algorithms into the building blocks for machine autonomy, interfacing them with industry standards.
The software is deployed edge-only on low-end silicon. Stabilizing vision and extracting motion to perceive in real-time. Machines building maps by simply recognising spaces as they pass through them. Weighing up information to make decisions.
Machines responding adaptably as in nature.
How it works
Our patented algorithms provide innate edge only autonomy for machines as a set of integrated software modules:
Passive omnidirectional vision (360x180 degrees 4π steradians), algorithmically stabilised at point of capture. Low-cost 2D CMOS cameras, high frame rates (150fps+) with no buffering.
Detects structure & motion in variable lighting, poor weather and aliasing without expensive active sensors or fragile feature extraction.
Vision-only collision prediction, avoidance & motion planning without huge 3D point cloud complexity and overhead.
Opteran Spatial Engine
Ultra-low memory (only 1kB/m2) facilitating edge-only dynamic learning and navigation indoors and outdoors with no data capture, labelling or training requirement.
Opteran Decision Engine
Natural brain decision making to action select and task-switch without using condition statements or reinforcement learning.
Advantages of Natural Intelligence algorithms
General purpose (air and ground), edge only, no network, no data, no training, secure and private, simple integration, standards-based (ROS1 / ROS2, Mavlink/PX4 or Ardupilot), ultra-low SWaP$.
Using low cost, low bandwidth colour vision:
|Colour||at sensor||not supported||not supported||at sensor||not supported|
|Distance||at sensor||at sensor||at sensor||central processing||not supported|
|Motion||at sensor||at sensor||at sensor||central processing||central processing|
|Localisation||at sensor||central processing||not supported||central processing||central processing|
|Bandwidth||Level 1 of 3||Level 3 of 3||Level 2 of 3||Level 3 of 3||Level 3 of 3|
|Cost||Level 1 of 3||Level 3 of 3||Level 2 of 3||Level 2 of 3||Level 3 of 3|
- At sensor
- Not supported
- Central Processing
Try the Opteran Mind
We provided several approaches to test our general purpose autonomy.
Our algorithms integrated with the Unreal Engine allow you to test the Opteran Mind in simulations that represent your business needs.
The Opteran Development Kit (ODK) with integrated software, hardware and pre-calibrated cameras allows you to integrate Opteran algorithms directly into your machine. Simply, plug in a USB or Ethernet to access PX4 over Mavlink or ROS APIs and get going.
To simplify this, we have pre-integrated the ODK into a drone and ground-based robot to allow you to test the Opteran Mind quickly and simply.