EP.30 THIS ROBOT WILL CLEAN UP THE MESS AFTER YOUR KIDS
New version of Atlas, robot that can navigate in space, a legged robot on wheels & much more ...
The robot that learned to move in weightlessness! πͺ
Meet SpaceHopper, a robot designed to navigate the challenging terrain of asteroids and moons. This robot is developed by a team of students from ETH Zurich and SpaceHopper. Its core functionality is its ability to jump and reorient itself mid-air, allowing it to traverse uneven surfaces easily.
The robot's custom testing rigs, including a counterweight system and gimbal, enable it to demonstrate its capabilities in a controlled environment. The team of nine students, hailing from various disciplines, worked together to design, engineer, and test SpaceHopper. SpaceHopper's impressive features include its ability to jump and reorient itself, making it an ideal candidate for asteroid and moon exploration. The robot's advanced capabilities are made possible by its custom-designed testing rigs, which simulate the harsh conditions found in space.
The project was made possible through the support of various sponsors, including Diamond, Platinum, Gold, Silver, and Patron-level partners. SpaceHopper is a testament to the power of collaboration and innovation, and its potential applications in space exploration are vast.
To infinity and beyond! π¨π½βπ β [read more here]
The king is dead, long live the king (Atlas)! π
Boston Dynamics has retired its hydraulic-powered Atlas humanoid robot and introduced an all-electric version. The new Atlas builds on decades of research and incorporates lessons learned from commercializing Spot and Stretch.
It features improved control policies, upgraded actuation, and minimized joint complexity, making it ideal for mobile manipulation in industrial settings. Safety and autonomy are critical components of the new Atlas. Boston Dynamics has developed advanced, functionally safe 3D vision and onboard systems to ensure safety. While Spot and Atlas are often teleoperated, the goal is to achieve greater levels of autonomy, enabling the robots to operate independently and make decisions without human intervention.
With investment flowing into humanoid companies and new players entering the market, the competition is heating up. The company will begin testing the all-electric Atlas with parent company Hyundai and select partners next year, starting with automotive production. As the dexterity of the robots improves over time, we can expect to see them performing a wide range of tasks and adapting to various facilities.
Their video already has 5 million views on YouTube! π
Mobile manipulation using quadrupeds π
Researchers have made a breakthrough in mobile manipulation using legged robots equipped with an arm, a concept known as legged loco-manipulation. The proposed framework, Visual Whole-Body Control (VBC), enables autonomous whole-body control using visual observations.
VBC consists of two policies: a low-level policy that tracks the end-effector manipulator position using all degrees of freedom, and a high-level policy that proposes the end-effector position based on visual inputs. Both policies are trained in simulation and transferred to real robots using Sim2Real technology. Extensive experiments have demonstrated significant improvements over baselines in picking up diverse objects in various configurations (heights, locations, orientations) and environments.
This innovative approach has the potential to revolutionize mobile manipulation, enabling robots to navigate complex environments and perform tasks with greater precision and flexibility. The ability to extend the workspace through whole-body control opens up new possibilities for robots to interact with their surroundings in a more human-like way.
No more cleaning up after your kid anymore! πΆπ» β [read more here]
Meme of the week π€
Contest: Best caption wins. Send yours via mail!
Deep reinforcement learning for humanoids! π¦Ώ
In this course, participants will delve into deep reinforcement learning for walking robots using MATLAB and Simulink's Robotics System Toolboxβ’.
Upon completion, learners will be able to:
Understanding the application of deep reinforcement learning (specifically DDPG algorithm) in controlling humanoid robot locomotion.
Learning how to set up, train, and evaluate reinforcement learning using Simulink models in Simscape Multibodyβ’ and Reinforcement Learning Toolboxβ’ environments.
Selecting appropriate states, actions, and reward functions for the reinforcement learning problem.
Designing and implementing neural network structures for training the control policy.
Universal Robots and Siemens partnership! π
Universal Robots (UR), the world's leading collaborative robot (cobot) company, has integrated the Standard Robot Command Interface (SRCI) into its software, enabling seamless connectivity with Siemens PLCs. This integration enhances the connectivity capabilities of UR's cobots, making it easier for customers to integrate them with Siemens PLCs.
The SRCI standardizes definitions and robot commands between UR cobots and Siemens PLCs, allowing for easier and quicker setup and deployment. The SRCI is available for UR's e-Series family and the next-generation robots UR20 and UR30, and can be installed and activated via add-on URcap software.
"This integration is a great advantage for our customers, allowing them to easily integrate and use UR's robots together with Siemens PLCs in their production." says Rolf Heinsohn, Senior Vice President, Factory Automation Segment Control at Siemens.
This breakthrough has the potential to accelerate factory automation and scale the use of robots in industry, making them simple and available to all employees.
Strong industrial partnership! π¦Ύ β [read more here]
A robot that rides on roller skates! πΌ
Autonomous wheeled-legged robots have the potential to transform logistics systems in urban environments, improving operational efficiency and adaptability. However, navigating urban environments poses unique challenges, requiring innovative solutions for locomotion and navigation.
Researchers have developed a fully integrated system that addresses these challenges, comprising adaptive locomotion control, mobility-aware local navigation planning, and large-scale path planning within the city. Using model-free reinforcement learning (RL) techniques and privileged learning, they developed a versatile locomotion controller that achieves efficient and robust locomotion over various rough terrains.
The controller seamlessly transitions between walking and driving modes, enabling effective navigation through challenging terrain and obstacles at high speed. The system was validated through autonomous, kilometer-scale navigation missions in Zurich, Switzerland, and Seville, Spain, demonstrating its robustness and adaptability.
Key contributions include:
Kilometer-scale autonomous deployments in urban environments
Intelligent local navigation with active exploration and obstacle negotiation
Terrain-adaptive hybrid locomotion with seamless transitions between walking and driving modes
Future of delivery robots? π β [read more here]
Keep reading with a 7-day free trial
Subscribe to we all are robots to keep reading this post and get 7 days of free access to the full post archives.