Boston Dynamics redesigns its trailer-unloading robot to cut downtime

Boston Dynamics introduced Stretch Gen 2 on November 3, 2026, rebuilding its autonomous trailer-unloading robot around a practical problem: keeping it working. The redesign replaces wiring, connectors and other components that contributed to downtime, while a more modular structure is intended to make repairs easier when something does go wrong.

The company says it analyzed more than 100,000 operating hours and tens of millions of boxes unloaded by first-generation Stretch robots. That experience guided changes aimed at the most common causes of unplanned stoppages. Rather than focusing only on how quickly the arm moves boxes, the announcement details how the machine withstands vibration, avoids mechanical stress and gets back into service.

Less wiring, fewer hard stops

One of the largest physical changes is inside the robot. Boston Dynamics says it removed 81% of the internal cabling—about half a mile of wire. The company identifies wiring as its biggest source of intermittent electrical faults and wear, making that reduction a central part of the reliability redesign.

• Connectors: Heavy-duty, automotive-grade connectors are designed to withstand vibration from rapid, high-torque box handling. They also address the vibration the mobile robot encounters while driving across uneven floors, according to the company.
• Gripper: New valve systems and more precise sensing change the component that grasps and places boxes. Boston Dynamics says those upgrades improve handling during fast-paced picks, targeting both the pickup and placement portions of the task.
• Joints: Continuous-rotation joints can move beyond 360 degrees without reaching a hard stop. The company says the previous stops caused mechanical stress; the new design removes that particular source of strain rather than merely reinforcing it.

Falling boxes prompted exterior changes, too. The emergency-stop control now sits lower and is recessed to reduce accidental activation. New panel materials are intended to make a falling box less likely to damage the exterior or trigger an unintended safety stoppage. Those changes target interruptions caused by impacts, not just failures inside the machine.

Repairs move closer to the warehouse team

Stretch Gen 2 also changes how technicians reach and replace parts. Its modular design supports what Boston Dynamics calls plug-and-play part swaps. Dynamic cables and complex wiring harnesses are gone, while standardized high-strength fasteners reduce tool changes. The company says these choices shorten hands-on repair work, alongside the effort to reduce how often service is needed.

Onboard diagnostics and maintenance tools are intended to let customers’ own maintenance teams perform more routine service. Boston Dynamics presents this as a way to reduce reliance on outside support. The service changes therefore cover both the physical work of replacing components and who can handle routine maintenance at the dock.

The unloading rate comes with conditions

Boston Dynamics says Stretch Gen 2 can unload up to 1,000 cases per hour. That is a peak claim, not a promised rate for every trailer: actual throughput depends heavily on case weight and how neatly or irregularly boxes are stacked. The company identifies multipick as the feature that pushes performance toward the peak.

The multipick feature allows the robot to handle multiple boxes in one picking cycle, as Interesting Engineering describes. That mechanism is distinct from simply repeating individual picks faster: several boxes can move during the same cycle, helping the robot reach higher unloading rates.

For power, the company says the battery lasts two shifts and a fast charger reaches 90% charge in about 75 minutes. Onboard lighting supports work in dark trailers. Its pretrained vision system, Boston Dynamics says, recognizes most packages from the first day without requiring programming.

The redesign also includes a new onboard computing platform. Boston Dynamics says regular software releases can improve case detection, motion planning, picking strategies and world modeling—the robot’s representation of its surroundings—across a fleet without hardware changes. Those updates provide a separate route to improving deployed robots: changing how they perceive and plan their work, rather than replacing physical components.