Machine Learning Technologies used in Robotics

By Jyoti Nigania |Email | Jul 16, 2018 | 5520 Views

Robotic systems consist of object or scene recognition, vision-based motion control, vision-based mapping, and dense range sensing, and are used for identification and navigation. As these computer vision and robotic connections continue to develop, the benefits of vision technology including savings, improved quality, reliability, safety, and productivity are revealed.

What are some most widely used machine learning technologies in robotics?

Answered by Mats Bauer on Quora.
Machine learning (ML) is gaining populism in all areas of technology, especially the field of robotics. Let me give you the most distinct and important examples for ML in robotics:

  • Computer Vision: In this field of robotics, machine learning is used to improve and adapt camera based vision. Areas, where these intelligent camera systems are used are for example autonomous driving, intelligent drones, smart video surveillance (automatically calling police when people are fighting), smart home camera systems (opening the door when you are approaching). As it is impossible to teach a Tesla every possible situation in traffic, the solution is to create an intelligent system, in which one Tesla learns to handle a new situation and teaches it to all others. This is what ML in computer vision is all about.

  • Assistive and imitation learning: This is a new arising field in robotics, in which we teach or use robotics in every day use. Assistive robotics means, that we take the robot arm by the hand and show them what to do. This is helpful, when you are too shaky to do tiny work or to weak to lift heavy items. The imitation learning is based on this, it is when you teach your robot how to move, and it repeats this movement after you. This way of supervised learning is great for frequent changes in movement, as it only takes a few minutes to teach the robot a new routine. Imitation learning is closely related to observational learning, a behavior exhibited by infants and toddlers. 
Imitation learning is also an umbrella category for reinforcement learning, or the challenge of getting an agent to act in the world so as to maximize its rewards. Bayesian or probabilistic models are a common feature of this machine learning approach. Imitation learning has become an integral part of field robotics, in which characteristics of mobility outside a factory setting in domains like domains like construction, agriculture, search and rescue, military, and others, make it challenging to manually program robotic solutions.

  • Medical Robotics: A very important field in new robotics, as it will enable more people to access medical help faster. Again, these robots will often be confronted with new situations, so learning and adapting is crucial. This is where ML comes in, to help analyse and predict situations based on learned data. An assistive robot is a device that can sense, process sensory information, and perform actions that benefit people with disabilities and seniors. Movement therapy robots provide a diagnostic or therapeutic benefit. Both of these are technologies that are largely still confined to the lab, as they're still cost-prohibitive for most hospitals in the U.S. and abroad.

In summary, ML in robotics is crucial when it comes to advancing the processes and improving, based on collected data. Only robots that are learning are robots with a future.

Source: HOB