bots_alive Teaches Robots to Solve Obstacle Courses

Brad Knox wants to design robots that are entertaining and affordable. The former MIT Media Lab engineer is running a Kickstarter campaign for bots_alive, artificial intelligence robots that can interact with each other and create movement paths.

The bots_alive platform uses a Hexbug Spider remote control toy, and then adds machine learning and multiple control interfaces. A picture is taken of the full field to give the bots_alive app the lay of the land. Next an infrared ‘blaster plugs into the user’s phone, creates paths for the bot to walk around the system. Several paths are generated to move around the obstacles (called the ‘learning blocks’) and reach an objective, so the exact movements of the bot are unpredictable. The example in the campaign video creates ten possible paths for the bot to take.












Adding the AI decal is the only modification that needs to be made to the Hexbug Spider device. The Hexbug batteries are said to show some drop in performance after two solid hours of running and then die out after another few full hours of use. The AG13 battery replacements can be purchased from Hexbug directly or Knox has posted sources on the campaign for 100 batteries at a $10 pricepoint. One note about the system is that the infrared blaster uses the headphone jack, and iPhone 7 users will need to find their phone box and headphone adapter to use the bot system.

bots_alive is a great project that fosters robotics and machine learning education while being entertaining and modular. The project is funded by a National Science Foundation Small Business Innovation Research grant with a long term goal of “delivering thousands of sophisticated robot pets across the world”. Knox has written several articles about machine learning and artificial intelligence in the past five years, and continues to update progress on his website and twitter. The campaign ends on February 16 and first units are expected to ship in September.