The field of artificial intelligence is rapidly expanding its reach into physical, real-world environments, a category often referred to as "physical AI." Recent advancements have demonstrated robots' remarkable capabilities in dynamic and competitive settings, from mastering complex sports like table tennis to completing challenging endurance races. These developments highlight a significant leap in AI systems that can perceive, decide, and act with speed and precision in unstructured environments.

A prime example of this progress is "Ace," an autonomous table tennis robot developed by Sony AI. Designed for a competitive sport demanding rapid decision-making and precise motor control, Ace has successfully competed against and defeated high-level human players in regulated matches. The robot's sophisticated architecture combines high-speed perception systems with AI-driven control, enabling it to execute complex shots under match conditions. Its trials, conducted under International Table Tennis Federation rules and officiated by licensed umpires, documented wins against elite players as early as April 2025, with subsequent victories over professional players reported in late 2025 and early 2026. This achievement is particularly notable given that previous table tennis robots, existing since the 1980s, could not match the performance of advanced human players. Peter Dürr, director at Sony AI Zurich and lead of the project, emphasized the challenge, stating, "Unlike computer games, where prior AI systems surpass human experts, physical and real-time sports like table tennis remain a major open challenge."

The technical complexities of table tennis, including the speed and variability of the ball, complex spin, and changing trajectories, necessitate rapid sensing and coordinated movement within tight time constraints. To address these, Ace's architecture integrates nine synchronized cameras and three vision systems that meticulously track the ball's movement and spin. This system processes visual data at a speed that is difficult for the human eye to resolve, capturing motion that would appear as a blur to humans. The robotic platform itself is equipped with eight joints: three for precise positioning, two for orientation, and three for managing shot force and speed, a configuration designed to meet the minimum mechanical requirements for competitive play. Uniquely, Ace was trained entirely in simulation, allowing it to develop its own strategies without human demonstration. This approach resulted in play patterns that differ from human opponents, making it unpredictable for players like Mayuka Taira, who remarked on the robot's lack of visible cues or emotional signals. Elite player Rui Takenaka also noted its strong handling of complex spins, though finding it more predictable on simpler serves. Sony AI's ongoing work focuses on improving Ace's adaptability during matches, with the perception and control techniques developed potentially applicable to broader fields such as manufacturing and service robotics.

Further showcasing the strides in physical AI, humanoid robots competed in the 2026 Beijing E-Town Humanoid Robot Half Marathon. This event saw more than 100 robots, alongside approximately 12,000 human participants on separate tracks, traversing a 21-kilometer course. A standout performer was "Lightning," a robot developed by Honor, which completed the race in an impressive 50 minutes and 26 seconds. This time was notably faster than the 57 minutes and 20 seconds recorded by Olympic runner Jacob Kiplimo at the Lisbon Half Marathon. Despite a collision with a barricade during the race, Lightning persevered to finish first, with other Honor robots securing second and third places. The significant improvement over the previous year's fastest robot, which completed the course in two hours, 40 minutes and 42 seconds, underscored the rapid progress in humanoid robotics. Organizers stated the event's purpose was to rigorously test humanoid robots in large-scale, real-world conditions, emphasizing autonomous navigation over remote control for official recognition. Honor engineers anticipate that technologies developed for Lightning, such as advanced structural reliability and liquid-cooling systems, will find practical applications in industrial scenarios.

These concurrent achievements in table tennis and long-distance running underscore the growing capabilities of physical AI. By demonstrating mastery over complex motor skills, rapid decision-making, and sustained endurance in dynamic, real-world settings, these robots are paving the way for advanced automation and intelligent machines that can operate effectively alongside or in support of humans across various industries and daily life.