Boston Dynamics' New Atlas: Humanoid Robot Built to Fall and Get Back Up

Like humans, robots learn from falling and rising.

Boston Dynamics' creative marketing unveiled the new Atlas, the robotics community's biggest news. Nearly five million people have seen the launch video, demonstrating the company's ongoing appeal.

Pras Velagapudi, the company's new Chief Technology Officer, stressed the value of learning experiences in developing robotics, saying unexpected situations are essential to real-world implementation and noted that in actual scenarios "You're going to see some falls," but each fall teaches robots to navigate various settings.

According to Boston Dynamics CTO Aaron Saunders, the company's fall policy goes against concerns about robot damage. As per TechCrunch, the company pushes robots to failure, studies them, and improves resilience to development. The electric Atlas's unusual ability to rebound from prone postures has real-world applications.

The Importance of Robots Getting Back Up After Falling

Boston Dynamics CEO Robert Playter stressed the necessity of robots' fall recovery, especially in dynamic contexts like factories and warehouses. Playter stressed that robots must be able to take falls and recover swiftly to minimize disturbances to automated activities, citing Spot.

Another major robotics company, Agility Robotics, agreed that falls would improve humanoid robots. Jonathan Hurst, the co-founder of Agility Robots, emphasized the significance of arms in preventing falls and aiding recovery, a concept that Boston Dynamics also incorporated into their bipedal robot approach.

Both businesses use reinforcement learning to teach robots to recover from unexpected falls. Integrating humanoid robots into operations necessitates agility, as disruptions can have major operational consequences.

This development comes as Chinese robotics firm Astribot unveiled the S1, which can travel 10 meters per second and carry 22 pounds per arm. The company's video shows the robot opening and pouring wine, shaving a cucumber, flipping a sandwich, and painting calligraphy.

According to Astribot, S1 accomplishes these tasks through imitation learning. This method trains the robot by exposing it to massive volumes of human behavior data, per Interesting Engineering.

Rising Conerns Over Robots

Astribot has not disclosed the S1's training or autonomous movement, but the video shows its capabilities. According to its website, Astribot created the S1 in 2022 and built it in a single year. The robot, according to their website, is "closest to human operating performance" and will undergo testing before release in 2024.

Amid the innovations in the field of robotics comes also rising concerns. Privacy groups and legal experts in New York worry about the Knightscope K5 robot assisting human cops, as per a USA Today report.

Robots like the K5 gather information and provide protection, but detractors say they don't reduce crime. According to legal expert Andrew Ferguson, these robots may be "security theater," providing a showy but ineffectual answer.

Privacy encroachment, weaponization, and unregulated robotic patrols are the biggest concerns. Previous public safety robot incidents tarnished public opinion. Critics pointed to body temperature checks at homeless shelters and border patrol assistance.

The Surveillance Technology Oversight Project executive director, Albert Fox Cahn, challenges robots' needs. He advises caution before adopting this technique due to its high cost and unproven efficacy.

ⓒ 2024 TECHTIMES.com All rights reserved. Do not reproduce without permission.
Join the Discussion
Real Time Analytics