Robot, which brings beer from the refrigerator (video)
German engineers have taught the humanoid robot TIAGo to supply the beer with the desired brand. The assistant himself will find a fridge and easily distinguish the bank bought from the nearest supermarket from a bottle from a well-chosen craft brewery. And since the project participates in the competition NVIDIA Jetson Developer Challenge, for their work team can get quite a financial reward.
How it works?
Developers connected to TIAGo a modular supercomputer NVIDIA Jetson TX2, which is responsible for "vision". Using the built-in camera with a depth sensor, a special algorithm scans the surrounding space, recognizes objects and classifies them (by shape, size, etc.). The robot itself calculates the route to the refrigerator, finds the handle and opens the door. Among the products on the shelves, he allocates beer bottles, picks up the necessary, closes the refrigerator and returns to the user.
According to the authors, for the training of the system they needed 60 images of bottles and 20 different variants of the background, and the whole process takes about 2 hours. For more accurate work of the algorithm it is possible to increase the sample, and then the information processing will be delayed by about 5-8 hours.
This is something new?
Not really. Robots Boston Dynamics has long learned to bring beer to the owner, and CES 2018 was introduced home assistant Aeolus , who also cope with this task. There are a lot of similar projects, but more often than not machines are lost when the right beverage is not in the right place. TIAGo with the "brain" of NVIDIA is more flexible in this regard.
What is the NVIDIA contest?
In October last year, the company launched a competition among engineers who use NVIDIA Jetson TX2 in their projects. All interested could apply for participation until February 18, 2018. In March, the finalists will be announced, all of whom are promised to be invited to the GTC 2018 computer conference in the Silicon Valley.
The winners will receive a cash prize of $ 10,000, a video card from NVIDIA TITAN Xp, another Jetson TX2 supercomputer in their collection and access to a deep training course.
Source: ChallengeRocket
