WIRED (FEB 2018) - If you want to teach a robot to pick up an object, you could do it the classical way and program it with line after line of code. Or like Kindred says its system works, you can use more modern approaches in artificial intelligence: reinforcement learning and imitation learning.
WIRED (DEC 2017) - At a San Francisco startup called Kindred, for example, engineers are teaching robots to do that final step of fulfillment. Using a technique called imitation learning, engineers steer the robot to show how best to grasp a wide range of objects you’d find at a marketplace like Amazon. “Some are soft and squishy, some are hard, some are heavy, some are soft,” says George Babu, co-founder of Kindred. “And there's no way you can program that.”
HARVARD BUSINESS REVIEW (JUL 2017) - Erik Brynjolfsson and Andrew McAfee report on the potential of AI, its practical implications, and the barriers to its adoption. The article notes a small but growing area within the field isreinforcement learning, highlighting robots created by Kindred that use machine learning to identify and sort objects they’ve never encountered before, speeding up the “pick and place” process in distribution centers for consumer goods.
MIT TECHNOLOGY REVIEW - The robots use machine-learning algorithms to operate, but if one runs into problems—say, while grasping and placing items inside a warehouse—the human can temporarily take control, and the robot will improve its performance via reinforcement learning.