Drones navigate unseen environments with liquid neural networks

MIT News April 19, 2023 Autonomous robots can learn to perform visual navigation tasks from offline human demonstrations and generalize online and unseen scenarios within the same environment they have been trained on. It is challenging for these agents to take a step further and robustly generalize to new environments with drastic scenery changes that they have never encountered. Researchers at MIT have developed a method to create robust flight navigation agents that successfully perform vision-based fly-to-target tasks beyond their training environment under drastic distribution shifts. They designed an imitation learning framework using liquid neural networks, a brain-inspired class of […]

Robots predict human intention for faster builds

Science  Daily April 5, 2023 To focus on enabling robots to proactively assist humans in assembly tasks by adapting to their preferred sequence of actions researchers at the University of Southern California proposed learning human preferences from demonstrations in a shorter, canonical task to predict user actions in the actual assembly task. The proposed system used the preference model learned from the canonical task as a prior and updates the model through interaction when predictions are inaccurate. They evaluated the proposed system in simulated assembly tasks and in a real-world human-robot assembly study and showed that both transferring the preference […]

DARPA’s Robotic In-Space Mechanic Aces Tests, on Track for Launch

DARPA  November 8, 2022 DARPA is seeking to create a persistent operational dexterous robotic capability in geosynchronous Earth orbit to enable on-orbit satellite repair and upgrade, extending satellite life spans, expanding the capabilities of existing satellites, enhancing spacecraft resilience, and improving the reliability of the current U.S. space infrastructure. All component-level tests are complete on DARPA’s Robotic Servicing of Geosynchronous Satellites (RSGS) program and the on-orbit demonstration mission is on schedule for launch in 2024. Following a period of checkout and calibration activities, the program anticipates on-orbit satellite servicing activities will begin in 2025. RSGS is intended to remain in […]

Scientists develop ‘mini-brains’ to help robots recognize pain and to self-repair

Science Daily  October 15, 2020 Sensory information processing in robot skins currently rely on a centralized approach where signal transduction is separated from centralized computation and decision-making, requiring the transfer of large amounts of data from periphery to central processors. An international team of researchers (Singapore, Italy) took a decentralized approach where intelligence is embedded in the sensing nodes, using a unique neuromorphic methodology to extract relevant information in robotic skins. They proposed a system to address pain perception and the association of nociception with tactile perception to trigger the escape reflex in a sensitized robotic arm. The system comprises […]

Researchers create fly-catching robots

TechXplore  September 28, 2020 An international team of researchers (Austria, Germany) has created a soft robot and demonstrated a series of simulation-guided lightweight, durable, untethered, small-scale soft-bodied robots that perform large-degree deformations at high frequencies up to 100 Hz. They are driven at very low magnetic fields down to 0.5 mT and exhibit a specific energy density of 10.8 kJ m−3 mT−1. They observed asynchronous strongly nonlinear cross-clapping behavior of the robots in experiments and analyzed by simulation, breaking ground for future designs of soft-bodied robots. The robots walk, swim, levitate, transport cargo, squeeze into a vessel smaller than their dimensions and can […]

Welcome to robot university (only robots need apply)

MIT Technology Review  November 7, 2019 Researchers at UC Berkeley are creating RoboNet, a database similar to the ImageNet created by Princeton University, consisting of annotated video data of robots in action. The trick is to have countless hours of video to learn from. They start by recording the way a robot interacts with, say, a brush to move it across a surface. Then they take countless hours of video to learn from many more videos of its motion and use the data to train a neural network on how best to perform the action. Once a robot has mastered […]

Flexible yet sturdy robot is designed to “grow” like a plant

MIT News  November 7, 2019 In a factory or warehouse the robots have a harder time winding through narrow spaces to carry out tasks such as reaching for a product at the back of a cluttered shelf etc. Engineers at MIT have developed a robot designed to extend a chain-like appendage flexible enough to twist and turn in any necessary configuration, yet rigid enough to support heavy loads or apply torque to assemble parts in tight spaces. When the task is complete, the robot can retract the appendage and extend it again, at a different length and shape, to suit […]

Autonomous quadruped designed to team with Soldiers

EurekAlert  September 24, 2019 In collaboration with universities and industry U.S. Army Combat Capabilities Development Command’s Army Research Laboratory developed the Legged Locomotion and Movement Adaptation (LLAMA), an autonomous quadruped mobility research platform system patterned after a working dog and similar animals. It works in concert with soldiers, lighten physical workloads, and increase mobility, protection and lethality. The all electric system has high torque actuators and algorithms for advanced perception, intelligence and control for autonomy and teaming. It is designed for mobility in structured and unstructured environments. The readiness of the platform is dependent on its mission, a logistics mission […]

Shape-shifting robots built from smarticles could navigate Army operations

EurekAlert  September 18, 2019 Working under a U.S. Army project a team of researchers in the US (Georgia Institute of Technology, Northwestern University) built robots entirely from smaller robots known as smarticles, unlocking the principles of a potentially new locomotion technique. The 3D-printed smarticles can do just one thing: flap their two arms. But when five of these smarticles are confined in a circle, they begin to nudge one another, forming a robophysical system known as a “supersmarticle” that can move by itself. The devices can change their location only when they interact with other devices while enclosed by a […]

Drones will fly for days with new photovoltaic engine

Tech Xplore  July 26, 2019 A team of researchers in the US (Laurence Berkeley National Laboratory, UCBerkeley, National Renewable Energy Laboratory, Swarthmore College, University of Colorado, Caltech) used the rear mirror to create thermophotovoltaic systems with unprecedented high thermophotovoltaic efficiency. This mirror reflects low-energy infrared photons back into the heat source, recovering their energy. Therefore, the rear mirror serves a dual function; boosting the voltage and reusing infrared thermal photons. This allows the possibility of a practical >50% efficient thermophotovoltaic system…read more. Open Access TECHNICAL ARTICLE