Researchers at Cornell University are working on a robot that can be programmed by voice command.
The machine is being developed by Professor Ashutosh Saxena and doctoral students Dipendra Misra and Jaeyong Sung, according to CNET. The research team is looking to teach robots to understand basic English speech from different voices, as well as fill in missing information and adapt to their environment.
The robot, designed in the Robot Learning Lab at Cornell University, includes a 3-D camera built with computer vision software. It uses the camera to scan its environment and identify the objects in it, Business Standard reported.
The team trained the robot to associate objects with their capabilities. For example, a pan can be poured into a form, and stoves can have objects put on top of them and can heat things. The robot has the ability to recognize the pan, locate the stove and water faucet and include the information in its actions. Telling the robot to "heat water" will result in it using the stove or microwave. The team said that if a person removed the pan or moved the robot to a different kitchen, it would still be able to carry out the same actions the next day.
Techniques called "machine learning" were used to train the robot to associate commands with flexible actions, Business Standard reported. Animated video simulations and recorded voice commands from different speakers are given to the machine.
The researchers tested the robot by giving it instructions to prepare ramen noodles and make the Italian desert affogato. The robot correctly performed the tasks 64 percent of the time, which the researchers said was three to four times better than methods previously used. The robot was able to fulfill the requests even when researchers altered the commands and environment while also filling in missing steps.
The team has started a crowdsourcing campaign for the robot, for which supporters can sign up on the Tell Me Dave website, CNET reported.
"With crowdsourcing at such a scale, robots will learn at a much faster rate," Saxena said.