Agility's Digit 5 trades grace for uptime

Jonathan Hurst spent more than a decade building legged robots that move like animals, and the humanoid his company just shipped is, in his own words, heavy, bulky, and missing the natural motion of its predecessors. His essay in IEEE Spectrum on Digit 5 is one of the more candid accounts of what commercializing a humanoid actually costs, and most of that cost has nothing to do with walking.

Agility's earlier research robot Cassie sold 10 units at $250,000 each, and that saturated the market. Then the early Digit deployments, Hurst writes, took so much effort to keep running that they arguably wiped out whatever profit they made. That is a deployment economics problem, not a locomotion problem. Digit 5's headline specs were defined around this prior experience. Agility says Digit 5 lifts 23 kg bins onto high shelves and runs 90 minutes on a 9-minute charge. If that runtime autonomy holds across a real shift, it means the robot is stopped for charging roughly one-tenth of the time, which matters more to a warehouse than the fluidity of the robot's gait.

For anyone evaluating humanoids, this is a great real world example of the challenge builders and integrators face. Industry wants a robot that operates like a forklift fleet. Uptime, service, and autonomy, are all far more important than fluid natural motion if it doesn't help do the job. Agility's factory can build more than 10,000 units a year, but capacity doesn't equal orders. Agility claims that Digit 5 is the first humanoid to work safely and economically at scale. It is a nice claim but what I want to see is real-world field reliability data from a few hundred units running real shifts.