AI Athletes: Robots Enter Tennis & Marathon Running | 2026 Advances
A humanoid robot has defeated its creator in a game of tennis, marking a significant milestone in the development of artificial intelligence and robotics. The achievement, demonstrated on Monday, March 23, 2026, by researchers utilizing the newly developed LATENT system, showcases a leap forward in robotic athleticism and dynamic learning capabilities.
The LATENT system, unveiled this week, enables humanoids to learn complex tennis maneuvers in real-time, crucially, by analyzing and adapting to “imperfect” human movement data. Previous attempts at creating athletic robots were hampered by their inability to respond to the unpredictable nature of sports, such as varying ball trajectories and uneven playing surfaces. According to experts, this new approach represents a “quantum leap” in the field.
“Instead of following a fixed script, the robot now understands the intent of a shot and dynamically adjusts,” explained a source familiar with the project. The AI breaks down movements – like a backhand or lateral step – into components that can be learned, allowing the robot to correct its swing and balance mid-motion. This capability extends beyond sports, with potential applications in manufacturing and rescue operations.
The advancement in robotic tennis follows closely on the heels of another landmark event in athletic robotics. On March 20, 2026, bipedal robots participated as official pacers in a marathon held in Xiamen, China. Their role was to maintain a consistent pace for human runners, offering unwavering endurance and consistency. Participants reported the robotic pacers were a “great motivational factor,” particularly in family-oriented runs. Organizers plan to allow robots and humans to share running lanes for the first time at the Beijing Yizhuang Half Marathon in April, a move previously prohibited for safety reasons.
These successes build upon rapid progress in robotic soccer, where models like the KAIST Humanoide v0.7 now demonstrate fluid and natural movements, including weight shifting, balance recovery, and rapid response to changing conditions. This progress is driven by training in virtual simulations, which are then translated into real-world performance. Newer models are reportedly running almost 180 percent faster and recovering from falls in significantly less time.
The development of athletic robots is fueled by a substantial economic incentive. The global robotics market is projected to exceed 100 billion euros by 2028, with governments in China and the United States viewing “embodied AI” as critical to national competitiveness. Analysts suggest that demonstrating robotic capabilities in the challenging environment of sports is a key step toward commercialization. “If a robot can navigate the chaos of a soccer game, it can likely navigate a full warehouse,” one market observer noted. Challenges remain, however, including extending battery life and ensuring safe navigation in complex environments.
Looking ahead, international sports federations are planning “cooperation matches” between human and robotic teams within the next 12 to 18 months. The goal is to foster trust and demonstrate the potential for humanoids to be safe and reliable partners. By 2027, robots are expected to be capable of completing marathons autonomously. Although a robot challenging a professional tennis player remains years away, the recent developments confirm that the mechanical athlete is no longer a futuristic concept, but a rapidly evolving reality.