JetsonHacks

Autonomous Deep Learning Robot

The Autonomous Deep Learning Robot is a TurtleBot 2 compatible robot which uses a NVIDIA Jetson TK1 Development Kit for its processor. The company Autonomous recently began delivery of the product, and uses the Jetsons TK1 GPU for deep learning acceleration.

Deep Learning Robot by Autonomous

Simon Birrell over at Artificial Human Companions recently wrote a review of the device which includes extra information about how to configure the robot from someone who has just unpacked one. Birrell writes

It’s a great deal – but the instructions are sparse to non-existent

As with other TurtleBot 2 machines, a Kobuki base from Yujin Robot is used, along with an ASUS Xtion Pro Live and a large bluetooth speaker. The robot is shipped as a kit, with three platforms and standoffs for the user to assemble. The robot is similar to the JetsonBot series that we ran last year with a Kobuki replacing the JetsonBot iCreate base. A big plus for the Deep Learning Robot is that all the parts come together in one big box, ready to assemble. Assembly should be straightforward with the additional information in the review article.

Lots of application software comes installed on the Deep Learning Robot, including ROS, Caffe, Torch, and Theano. In his article, Birrell discusses the first logon into the Jetson, along with setting up the network. Also discussed is the first startup of ROS and using teleoperation to move the robot around.

In the last part of the article, Caffe is tested for some simple image recognition tasks including the obligatory cute kitten test.

The Artificial Human Companions article is worth looking at if you’re curious about the Autonomous Deep Learning Robot. I noticed that there have been a few articles about the robot on that site, so go check it out.