Researchers from NVIDIA, led by Stan Birchfield and Jonathan Tremblay, developed a first of its kind deep learning-based system that can teach a robot to complete a task by just observing the actions of a human. The method is designed to enhance communication between humans and robots and at the same time further research that will enable people to work alongside robots seamlessly.
“For robots to perform useful tasks in real-world settings, it must be easy to communicate the task to the robot; this includes both the desired result and any hints as to the best means to achieve that result,” the researchers stated in their research paper. “With demonstrations, a user can communicate a task to the robot and provide clues as to how to best perform the task.”
Using NVIDIA TITAN X GPUs, the researchers trained a sequence of neural networks to perform duties associated with perception, program generation, and program execution. As a result, the robot was able to learn a task from a single demonstration in the real world.
Once the robot sees a task, it generates a human-readable description of the steps necessary to re-perform the task. The description allows the user to quickly identify and correct any issues with the robot’s interpretation of the human demonstration before execution on the real robot.
The key to achieving this capability is leveraging the power of synthetic data to train the neural networks. Current approaches to training neural networks require large amounts of labeled training data, which is a serious bottleneck in these systems. With synthetic data generation, an almost infinite amount of labeled training data can be produced with very little effort.
This is also the first time an image-centric domain randomization approach has been used on a robot. Domain randomization is a technique to produce synthetic data with large amounts of diversity, which then fools the perception network into seeing the real-world data as simply another variation of its training data. The researchers chose to process the data in an image-centric manner to ensure that the networks are not dependent on the camera or environment.
“The perception network as described applies to any rigid real-world object that can be reasonably approximated by its 3D bounding cuboid,” the researchers said. “Despite never observing a real image during training, the perception network reliably detects the bounding cuboids of objects in real images, even under severe occlusions.”
For their demonstration, the team trained object detectors on several colored blocks and a toy car. The system was taught the physical relationship of blocks, whether they are stacked on top of one another or placed next to each other.
In the video above, the human operator shows a pair of stacks of cubes to the robot. The system then infers an appropriate program and correctly places the cubes in the correct order. Because it takes the current state of the world into account during execution, the system is able to recover from mistakes in real time.
The researchers will present their research paper and work at the International Conference on Robotics and Automation (ICRA), in Brisbane, Australia this week.
The team says they will continue to explore the use of synthetic training data for robotics manipulation to extend the capabilities of their method to additional scenarios.
via NVIDIA: Read the research paper
Receive an email update when we add a new ROBOT LEARNING article.
The Latest on: Robot learning
via Google News
The Latest on: Robot learning
CVCompiler is an robot that fixes your resume to make you more competitive
on November 21, 2018 at 7:14 am
Until then, why not let a machine-learning robot tell you what you’re doing wrong in trying to get a job? That is, before it replaces you completely. […]
RoboTurk: A crowdsourcing platform for imitation learning in robotics
on November 21, 2018 at 6:31 am
Imitation learning is a branch of machine learning that trains machines to mimic human behavior while completing particular tasks. These techniques show great promise in the field of robotics, as they ... […]
Norwegian Startup Wants To Eradicate Loneliness With Robots
on November 21, 2018 at 3:45 am
"Another particular success was the fact that in May 2017, the Norwegian Ministry of Education accredited No Isolation, confirming AV1 as a ‘valid option for reduced learning’ and stating that the use ... […]
RS Components introduces robot-based learning to students and educational establishments
on November 20, 2018 at 10:59 pm
RS Components (RS), the trading brand of Electrocomponents plc (LSE:ECM), a global multi-channel distributor, announced the availability of a new product, Sphero’s SPRK+, which is designed to provide ... […]
Recode Daily: Learning to love robots
on November 20, 2018 at 4:07 am
The stocks of Amazon, Apple, Facebook, Google and Netflix have all declined 20 percent or more in value in recent months, pushing them into bear market territory. The Dow Jones Industrial Average ende... […]
From Caregivers to Hospital and Surgery Assistants: The Future of Healthcare Robotics
on November 19, 2018 at 3:58 pm
Her research team has developed a number of methods to make this easier, focused on human-robot teaming, machine learning, and computer vision. “These methods are fast, don’t require large training se... […]
Enhancing learning with robots
on November 17, 2018 at 7:25 pm
Hong Kong (HKSAR) - A robot drama education programme, funded by the Innovation & Technology Fund for Better Living, will help children diagnosed with Autism Spectrum Disorder. News.gov.hk interviewed ... […]
Robots Finally Learning to Clean the Bathroom
on November 15, 2018 at 1:39 pm
A useful general home robot, as far as I’m concerned, needs to be able to do three things: fold laundry, wash dishes, and clean toilets. That’s all it would take to make me happy. We’ve seen ... […]
Explore Global Robotics Learning Software market report, trends, size, share, analysis, estimations and forecasts to 2023
on November 14, 2018 at 7:27 pm
The report gives a thorough overview of the present growth dynamics of the global Robotics Learning Software with the help of vast market data covering all important aspects and market segments. The r... […]
via Bing News