发布于

A quadcopter that works in the air and underwater and also has a suction cup for hitching a ride on a host

A team of researchers at Beihang University, working with colleagues at Imperial College London and Swiss Federal Laboratories for Materials Science and Technology, has developed a quadcopter drone that is capable of flying in the air and maneuvering underwater. It also has a suction cup for hitching a ride on a host. They describe their drone in the journal Science Robotics.
发布于

New method allows robot vision to identify occluded objects

When artificial intelligence systems encounter scenes where objects are not fully visible, they have to make estimations based only on the visible parts of the objects. This partial information leads to detection errors, and large training data is required to correctly recognize such scenes. Now, researchers at the Gwangju Institute of Science and Technology have developed a framework that allows robot vision to detect such objects successfully in the same way that we perceive them
发布于

Google researchers teach robots to learn by watching

end effectors

Different robot end effectors.

Roboticists usually teach robots new tasks by remotely operating them through performing a task. The robot then imitates the demonstration until it can perform the task on its own.

While this method of teaching robots is effective, it limits demonstrations to lab settings, and only programmers and roboticists can do the demonstrations. A research team at the robotics department at Google has been developing a new way for robots to learn.

Humans learn by watching all the time, but it’s not a simple task for robots to take on. This is difficult for robots because they look different than humans. For example, a robot with a two-fingered gripper won’t gain much knowledge about how to pick up a pen from watching a human with a five-fingered hand pick one up.

To tackle this problem, the team introduced a self-supervised method for Cross-Embodiment Inverse Reinforcement Learning (XIRL).

This method of teaching focuses on the robot learning the high-level task objective from videos. So, instead of trying to make individual human actions correspond with robot actions, the robot figures out what its end goal is.

It then summarizes that information in the form of a reward function that is invariant to physical differences like shape, actions and end effector dynamics. By utilizing the learned rewards and reinforcement learning, the research team taught robots how to handle objects through trial and error.

The robots learned more when the sample videos were more diverse. Experiments showed that the team’s learning method led to two to four times more sample efficient reinforcement learning on new embodiments.

The team has made an open-source implementation of its method and X-MAGICAL, its simulated benchmark for cross-embodiment imitation, to let others extend and build on their work.

X-MAGICAL was created to evaluate XIRL’s performance in a consistent environment. The program challenges a set of agent embodiments, that have different shapes and end effectors, to perform a task. The agents perform the tasks in different ways and at different speeds.

teaching

Demonstrating different shapes performing a task in X-MAGICAL. | Source: Google

The team also taught using real-world human demonstrations of tasks. They used their method to train a simulated Sawyer arm to push a puck into a target zone. Their teaching method also outperformed baseline methods.

The research team included Kevin Zakka, Andy Zeng, Pete Florence, Jonathan Tompson and Debidatta Dwibedi from robotics at Google, and Jeannette Bohg from Stanford University.

The post Google researchers teach robots to learn by watching appeared first on The Robot Report.

发布于

Soft robotic origami crawlers

Materials scientists aim to develop biomimetic soft robotic crawlers including earthworm-like and inchworm-like crawlers to realize locomotion via in-plane and out-of-plane contractions for a variety of engineering applications. While such devices can show effective motion in confined spaces, it is challenging to miniaturize the concept due to complex and limited actuation. In a new report now published in Science Advances, Qiji Ze and a team of scientists in mechanical engineering and aerospace engineering at Stanford University and the Ohio State University, U.S., described a magnetically actuated, small-scale origami crawler exhibiting in-plane contraction. The team achieved contraction mechanisms via a four-unit Kresling origami assembly to facilitate the motion of an untethered robot with crawling or steering capacity. The crawler overcame large resistances in severely confined spaces due to its magnetically tunable structural stiffness and anisotropy. The setup provided a contraption for drug storage and release with potential to serve as a minimally invasive device in biomedicine.