Atlas VLA Research Intern

US immigration agency ICE to deploy robot dogs. Updated August 29 2026 - 10:33am, first published 10:29am The controversial US Immigration and Customs Enforcement (ICE) agency is set to deploy robot dogs. Photo: AP PHOTO The controversial US Immigration and Customs Enforcement (ICE) agency plans to deploy dog-like robots under an anticipated deal with a leading global robotics company, according to government contract award documents. Massachusetts-based Boston Dynamics is expected to be awarded the $US2 million ($A2.8 million) contract before the end of this year, according to a recent notification by the US Department of Homeland Security (DHS). The remote-controlled robots are intended for situational awareness and risk assessment, according to the justification for the acquisition by the department, which oversees the activities of ICE. Their use will enhance the authorities' ability to respond to incidents "involving dangerous, confined, unstable, or difficult-to-access areas," it said. Nothing is known about any potential attack capability of the robot canines. The standard SPOT robots are fitted with a high-resolution 360-degree all-round camera, ultra-bright LED light and a light detection system. Deportation raids carried out by ICE officers are part of President Donald Trump's strict immigration policy. In January, the agency came under heavy criticism after federal officers shot and killed US citizens Renée Good and Alex Pretti on a public street in Minneapolis. There have since been other fatal incidents, for example in Houston in Texas and in the state of Maine. More recently, the agency also caused a stir with its planned procurement of electric-shock gloves. It is not unusual for authorities and security forces to make use of four-legged robots. Dog-like robots from the same company have already been deployed at the German port of Hamburg to inspect the safety of bridges. Similar robots can also be used in the event of bomb threats or other dangerous situations. Australian Associated Press

Humanoid robots learn agile moves from human motion data with ZEST. A new control framework called ZEST enables humanoid and quadruped robots to perform complex, agile movements by learning from human motion capture, video, and animation data, reducing the need for task-specific engineering Researchers at the RAI Institute and Boston Dynamics have introduced Zero-shot Embodied Skill Transfer (ZEST), a reinforcement learning framework designed to teach legged robots a broad range of agile, whole-body movements using human motion data. Unlike conventional approaches that require separate training or manual controller tuning for each new skill, ZEST enables robots to acquire diverse movements-including crawling, breakdancing, and backflips-through a single training phase conducted entirely in simulation. The system is intended to reduce the engineering burden associated with deploying new robotic behaviors and to expand the repertoire of actions available to humanoid and quadruped robots. ZEST is capable of learning from three primary sources: high-fidelity motion-capture recordings, single-camera video footage, and keyframe animation. Human movements are converted into robot-compatible trajectories using kinematic retargeting, while animation data can supply skills that are difficult or unsafe for humans to demonstrate. The framework does not require explicit labeling of ground contact points in the demonstration data, which simplifies the preparation of training material and allows for a wider variety of input sources. Simulation and real-world transfer. The core of ZEST is a feedforward reinforcement learning policy trained in simulation, relying solely on onboard proprioceptive sensors during deployment. The policy is transferred to physical robots without additional fine-tuning, avoiding the need for state estimators, long observation histories, or complex reward engineering. To address the challenge of sim-to-real transfer, the researchers incorporated simplified models of closed-chain actuators and accounted for actuator effects such as power limits, motor saturation, and friction. This approach aims to preserve essential dynamics while reducing simulation complexity. During training, ZEST uses adaptive sampling to focus learning on the most challenging segments of each motion trajectory. The system tracks failure rates for fixed-duration segments and increases sampling frequency for those with higher failure rates, while maintaining a minimum sampling rate for easier segments to prevent skill loss. An automatic curriculum is implemented using a virtual assistive force, which is gradually reduced as the policy improves, allowing the robot to attempt more difficult movements with less external support. Evaluation and limitations. ZEST was evaluated on three commercial robots: Boston Dynamics' Atlas humanoid, Unitree's G1 humanoid, and Boston Dynamics' Spot quadruped. Each policy required approximately 10 hours of training-about 7,000 iterations-on a single NVIDIA L4 GPU. On Atlas, the system reproduced a range of complex movements, including crawling, forward rolls, cartwheels, army crawling, and breakdancing. Video-derived skills included dancing, soccer kicking, and box climbing, while animation data enabled Spot to perform continuous backflips and barrel rolls. The demonstration was conducted in controlled, flat, and non-slippery environments, and the system has not yet shown generalization to entirely novel movements or unstructured terrain. While ZEST reduces the need for task-specific engineering, it remains limited by the quality and diversity of available motion data and by the constraints of the physical robots. The framework does not currently support adaptation to new environments or tasks without retraining, and its performance in the presence of obstacles, uneven surfaces, or unexpected disturbances has not been established. The researchers note that future work will focus on enabling zero- and few-shot adaptation, continual learning, and higher-level control interfaces, such as language or keyframe commands. Context and related work. The development of ZEST reflects a broader trend in robotics toward more flexible, data-driven skill acquisition. Previous research has often relied on extensive manual engineering or teleoperation to achieve complex behaviors, limiting scalability and adaptability. Recent advances in reinforcement learning and motion retargeting have enabled robots to learn from a wider range of human demonstrations, but challenges remain in transferring these skills to real hardware and in ensuring reliable performance outside controlled settings. For example, other research efforts, such as the LUMO robot, have explored adaptive mobility and interaction in varied environments, as discussed in this recent report on all-terrain humanoid navigation. As robots are increasingly deployed in dynamic and human-centered environments, the ability to acquire and execute complex movements safely and reliably will depend not only on advances in learning algorithms but also on robust sensing, actuation, and safety engineering. The ZEST framework represents a step toward more generalizable skill transfer, but its current limitations highlight the need for continued research on adaptation, robustness, and human oversight in real-world deployment. Reinforcement learning is a machine learning approach in which an agent learns to make decisions by interacting with an environment and receiving feedback in the form of rewards or penalties. In robotics, reinforcement learning can be used to train control policies that map sensor inputs to motor commands, enabling robots to acquire new skills through trial and error. However, transferring policies trained in simulation to physical robots-known as sim-to-real transfer-remains challenging due to differences between simulated and real-world dynamics. Techniques such as domain randomization, actuator modeling, and adaptive curricula are often used to bridge this gap, but reliable deployment still requires careful engineering and validation.

BuiltWorlds survey tracks contractors' marked uptick in robotics adoption. Sources: BuiltWorlds, Chicago; CP staff The BuiltWorlds 2026 Equipment & Robotics Benchmarking Report indicates a year-over-year doubling of jobsite robotics usage or testing by general and specialty contractors. Among respondents to the survey behind the report, 79 percent reported employing jobsite robotics to some degree; 32 percent noted they had "piloted or trialed" an automation solution on at least one jobsite, up from 12 percent in the 2025 survey. "These systems are engineered to operate within the dynamic, often unpredictable environments of active construction sites, improving productivity, worker safety, task consistency and overall operational efficiency," says BuiltWorlds Research Director Audrey Lynch, report author. "As construction projects grow larger, more complex, and increasingly data-driven, demand for reliable, high-performing automation continues to rise." More specifically, 2026 survey data shows that while reducing manual effort - a factor among 63 percent of respondents - and addressing safety concerns (56 percent) continue to be key benefits driving adoption, the ability to improve accuracy was the most cited factor (75 percent) in contractors turning to robotics. The accuracy element suggests contractors increasingly value the precision and quality gains robotics deliver, alongside labor and safety benefits, notes Lynch. While survey respondents identified a range of robotic solutions on jobsites the world over, FieldAI, Boston Dynamics and Dusty Robotics stand out for ratings and usage, she adds: "FieldAI because it was the most highly rated; BostonDynamics because it was the most adopted; and Dusty because it was the most piloted." FieldAI and Boston Dynamics rank in the On-Site category of BuiltWorlds' just-released 2026 Robotics Top 50. On-Site spans Inspection & Monitoring, Installation & Finishing and Layout subcategories. Alongside On-Site are Automated Machinery & Related Solutions, with Demolition, Earthmoving & Excavating and Material Transport subcategories, plus Offsite Robotics, with 3D Printing & Additive Manufacturing and Prefabricated Component Manufacturing subcategories. On-Site/Installation includes Advanced Construction Robotics, developer of the automated TyBot rebar tying technology; Offsite/3D includes Cobod and Icon, leaders in structural-grade concrete or mortar printing assemblies. "The companies that made this year's list show how global robotics have become," says Audrey Lynch. "The 2026 honorees include companies from 16 countries." While many showcased in this year's list operate out of the U.S., she adds, others are based in Austria, Canada, Denmark, Germany, Hong Kong, India, Japan, Lichtenstein, Netherlands, Norway, Singapore, South Korea, Sweden, Switzerland and the U.K.