Drones Gain the Power of Sense

Myriad 2 Chip Offers Vision and Avoidance Skills

Movidius, a company headquartered in California, develops silicon-based vision technology for connected objects. Google, for example, used the company’s Myriad 1 chip for Project Tango (see Google and Lenovo Prepare to Tango), an endeavour to bring smart sensing talents to tablets. Its goal was to build computing devices that have a thorough knowledge of their local environment.

Now Movidius has announced that its Myriad 2 chip are being on board a drone made by DJI. The $1,400 Phantom 4 drone can sense and avoid obstacles in real time. This type of advanced visual guidance is a first for a consumer-grade drone and provides an indication of where the market is heading for these products.

DJI, a leading maker of drones and aerial cameras, is developing drones that can be self-controlled, sensing and avoiding surrounding objects in the air in real time to prevent crashes and injuries. Drone makers and users are coming under increasing regulatory scrutiny given a series of near-misses with commercial aircraft.

We expect sensing chips such as the Myriad 2 to enable a broad range of new industrial and recreational applications that were not previously possible. Such chips are also expected to find applications in products such as security cameras and virtual reality headsets. Movidius will have competition from semiconductor bellwethers Qualcomm and Intel, as well as Nvidia. Silicon suppliers are gearing up to exploit opportunities in presented by the Internet of things, developing chips optimized for drones and other forms of robotics, with significant investment in computer vision and other forms of machine learning within devices.

This past January at International CES 2016, Qualcomm introduced Snapdragon Flight as well as other processor solutions for drones and robots, enabling a connected device to, as Qualcomm says, “see and sense its surroundings”. Intel announced a deal with Yuneec to embed its RealSense technology to be used in the latter’s Typhoon H drone. Nvidia has its Jetson platform used in robotics, which we covered in our report on Embedded World 2016 in Nuremberg. Sensors will be a key component in a new generation of autonomous moving objects. Chipsets are on the move and robots are now in the know.