The Oregon-based company said engineers conducted the test to evaluate Digit’s actuator strength, joint durability, and real-time control systems.

Deadlifts were selected because they require synchronized movement across the entire body. The task forces continuous adjustments in the arms, legs, and torso to keep the load centered and prevent tipping, placing higher demands on coordination than simpler motions.

Each increase in weight intensifies these demands, ensuring the robot operates beyond routine conditions while maintaining consistency without wobbling or resets. Integrating weight into the training process allows the system to model dynamic balance changes that static programming cannot replicate.

Thousands of simulated trials refine grip stability and posture control, enabling smooth transfer to real-world execution, according to tech site Techeblog.

Agility said the approach reflects a shift from manually programmed joint movements to adaptive, learned control systems. Digit can now detect weight distribution, respond in real time, and complete repeated lifting cycles autonomously.

Digit is a bipedal humanoid robot developed for repetitive industrial tasks, according to Interesting Engineering. The company said recent upgrades to the robot’s battery and overall design, including improved limbs and advanced end effectors, expand its manipulation range, enabling tasks such as stacking totes, loading carts, and handling varied materials. A streamlined production process is also expected to accelerate manufacturing and support broader adoption, it added.

In February, Agility said Toyota Motor Manufacturing Canada, the automaker’s largest production operation outside Japan, signed a commercial Robots-as-a-Service agreement following a successful pilot with Digit, Forbes reported.

It is also deploying Digit at GXO, which it described as the world’s largest pure-play contract logistics provider, German manufacturing company Schaeffler, and Amazon.