by Alain Tiquet

With only one week to go until the Emerging Companies Summit (ECS), taking place during the GPU Technology Conference, we wanted to continue our series preveiwing some of the great startups that will be presenting. Each ECS start-up is transforming industries through innovative, GPU-based technology, and Universal Robotics is raising the bar in the areas of artificial intelligence, 3D vision … and robots.

Here’s a sneak preview of what this Nashville, Tenn.-based company will be demo-ing at ECS:

Universal Robotics makes software that lets machines learn from interactions with the physical world and adapt to better perform specific tasks. The learning and adapting part is covered by Universal Robotics AI software called Neocortex.

Specifically developed to complete tasks that are never the same – like unloading a pallet of assorted boxes – Neocortex’s fitting tagline is, “Automating Intelligence.”  Just like a human brain, Neocortex needs sensory input. So Universal Robotics combines AI with a suite of sensors including which their 3D vision software called Spatial Vision – 3D vision positioning software that uses two web cameras to determine positions in space down to the millimeter. Spatial Vision is also available to those looking to create 3D movies, improve facial recognition, or virtually any other vision-based system.  The final piece is Universal’s Visual Servoing software, which controls machine movement, allowing a robot to react and move based on 3D vision.

The company uses NVIDIA Tesla GPUs and CUDA programming to run the Neocortex learning algorithms and to provide the processing power for 3D vision-dependent capabilities like robot movement and collision avoidance.

Equipped with Universal Robotics’ suite of software can be used for work that is too costly, difficult, or dangerous for humans to do. Applications could include hazardous waste cleanup or high-speed materials handling. In fact, Universal Robotics already has one partnership with robot manufacturer Motoman to provide the software for industrial robots. The underlying Spatial Vision software can be used to turn any two webcams into an accurate, cost-effective, and easy 3D vision system.