Agility Robotics, the creator of the bipedal robot Digit, is opening a manufacturing plant in Salem, Oregon that will give the company the capacity to produce more than 10,000 humanoid robots a year. The 70,000 square foot factory coined "RoboFab" is set to open later this year and will employ upwards of 500 workers in Salem. Agility Robotics says its facility will also employ its very own Digits, the iconic humanoid robot, in the new factory. The Digits will help move, load and unload warehouse goods. The company says some customers can expect delivery of the first Digits in 2024, with general market availability in 2025.

Sony has launched the "Aibo Foster Parent" program for its $2,900 robot dogs, allowing owners whose basic plans have been canceled to donate them. The company will then refurbish the Aibos as necessary and donate them to medical facilities, foster homes and other organizations. The aim is to "make Aibo more sustainable," the company said, by giving the units a second home where they can provide emotional support and more. The Aibo robot dogs in question are the newer 2019 ERS-1000 units still being sold, which are a reboot of its "entertainment" robotic dogs from the late 1990s. It's not exactly a fully charitable project, as Sony will charge "foster parents" an unnamed fee for service. It also notes that depending on condition, some donated units may serve strictly as parts for other Aibo robots.

Source: OpenAI's DALL·E 2 with prompt "a hyperrealistic picture of a robot reading the news on a laptop at a coffee shop" Welcome to the third edition of Robo-Insight, a biweekly robotics news update! In this post, we are excited to share a range of new advancements in the field and highlight progress in areas like motion, unfamiliar navigation, dynamic control, digging, agriculture, surgery, and food sorting. In a world of constant motion, a newly developed robot named M4 (Multi-Modal Mobility Morphobot) has demonstrated the ability to switch between eight different modes of motion, including rolling, flying, and walking. Designed by researchers from Caltech's Center for Autonomous Systems and Technologies (CAST) and Northeastern University, the robot can autonomously adapt its movement strategy based on its environment. Created by engineers Mory Gharib and Alireza Ramezani, the M4 project aims to enhance robot locomotion by utilizing a combination of adaptable components and artificial intelligence. Speaking of movement, researchers from the Hamburg University of Applied Sciences have presented an innovative navigation algorithm for a mobile robot assistance system based on OpenStreetMap data.

Getting a leg up from mobile robots comes down to getting a bunch of legs. Adding legs to robots that have minimal awareness of the environment around them can help the robots operate more effectively in difficult terrain, my colleagues and I found. We were inspired by mathematician and engineer Claude Shannon's communication theory about how to transmit signals over distance. Instead of spending a huge amount of money to build the perfect wire, Shannon illustrated that it is good enough to use redundancy to reliably convey information over noisy communication channels. We wondered if we could do the same thing for transporting cargo via robots.

Eight years ago, the Pentagon's Defense Advanced Research Projects Agency organized a painful-to-watch contest that involved robots slowly struggling (and often failing) to perform a series of human tasks, including opening doors, operating power tools, and driving golf carts. Clips of them fumbling and stumbling through the Darpa Robotics Challenge soon went viral. Today the descendants of those hapless robots are a lot more capable and graceful. Several startups are developing humanoids that they claim could, in just a few years, find employment in warehouses and factories. Jerry Pratt, a senior research scientist at the Institute for Human and Machine Cognition, a nonprofit research institute in Florida, led a team that came second in the Darpa challenge back in 2015.

The more legs a robot has the better it seems to be at travelling over rough terrain. Baxi Chong at the Georgia Institute of Technology and his colleagues built a range of multi-legged robots from 3D-printed body segments. Each segment had two legs and several motors. The robots had between six and 16 legs in total. None of the robots had any sensors or cameras so they couldn't see their environment and just moved in pre-programmed way that mimicked arthropods.

Exploration is a fundamental challenge in reinforcement learning (RL). Many current exploration methods for deep RL use task-agnostic objectives, such as information gain or bonuses based on state visitation. However, many practical applications of RL involve learning more than a single task, and prior tasks can be used to inform how exploration should be performed in new tasks. In this work, we study how prior tasks can inform an agent about how to explore effectively in new situations. We introduce a novel gradient-based fast adaptation algorithm - model agnostic exploration with structured noise (MAESN) - to learn exploration strategies from prior experience. The prior experience is used both to initialize a policy and to acquire a latent exploration space that can inject structured stochasticity into a policy, producing exploration strategies that are informed by prior knowledge and are more effective than random action-space noise. We show that MAESN is more effective at learning exploration strategies when compared to prior meta-RL methods, RL without learned exploration strategies, and task-agnostic exploration methods. We evaluate our method on a variety of simulated tasks: locomotion with a wheeled robot, locomotion with a quadrupedal walker, and object manipulation.

Bayesian optimization (BO) is a popular approach to optimize expensive-toevaluate black-box functions. A significant challenge in BO is to scale to highdimensional parameter spaces while retaining sample efficiency. A solution considered in existing literature is to embed the high-dimensional space in a lowerdimensional manifold, often via a random linear embedding. In this paper, we identify several crucial issues and misconceptions about the use of linear embeddings for BO. We study the properties of linear embeddings from the literature and show that some of the design choices in current approaches adversely impact their performance. We show empirically that properly addressing these issues significantly improves the efficacy of linear embeddings for BO on a range of problems, including learning a gait policy for robot locomotion.

In this paper we examine the practical use of hardware neural networks in an autonomous mobile robot. We have developed a hardware neural system based around a custom VLSI chip, EP(cid:173) SILON III, designed specifically for embedded hardware neural applications. We present here a demonstration application of an autonomous mobile robot that highlights the flexibility of this sys(cid:173) tem.

The problem that we address in this paper is how a mobile robot can plan in order to arrive at its goal with minimum uncertainty. Traditional motion planning algo(cid:173) rithms often assume that a mobile robot can track its position reliably, however, in real world situations, reliable localization may not always be feasible. Partially Observable Markov Decision Processes (POMDPs) provide one way to maximize the certainty of reaching the goal state, but at the cost of computational intractability for large state spaces. The method we propose explicitly models the uncertainty of the robot's position as a state variable, and generates trajectories through the augmented pose-uncertainty space. By minimizing the positional uncertainty at the goal, the robot reduces the likelihood it becomes lost.