Manipulation Ambition Compared With Logistics Deployment Evidence

Figure and Agility publish different types of industrial evidence. Figure reports automotive part handling and sequencing at BMW, while Agility reports tote-handling deployments and operating hours for Digit. The comparison keeps robot generation, task and site attached to every metric.

Figure 02 and Figure 03 are treated as separate generations, as are Digit 4 and Digit 5. A metric from an older generation is not assigned to the newer robot unless the company reports it for that version.

Figure combines humanoid mobility, hands and its Helix control stack for manipulation workflows. Digit combines bipedal mobility with purpose-built arms and grippers and is integrated into warehouse workflows through Agility’s fleet software.

For buyers, the useful comparison is the complete task loop: object presentation, robot motion, grasp or tote handling, recovery, resets, supervision and facility integration. Public sources do not expose all of those fields for both companies.

Figure publishes Helix demonstrations and partner announcements. Agility publishes logistics deployments and commercial agreements, though site-level uptime and intervention data remain limited. The systems are not treated as interchangeable.

Figure’s BMW record and Agility’s Digit record are kept by robot generation and task. Figure 02 production data is not assigned to Figure 03, and Digit 4 operating hours are not assigned to Digit 5. Missing intervention, reset and full-shift failure data remain undisclosed.

Figure reports more than 1,250 hours of Figure 02 operation at BMW and a later Figure 03 logistics sequencing workflow. Agility reports more than 65,000 operating hours for Digit 4, including a 100,000-tote milestone at GXO at about 98% on-task accuracy. These are company-reported records from different tasks and sites, so they should not be converted into a single performance score.

A stronger comparison would report the same task, shift length, throughput definition, intervention rate, reset policy and failure categories for both systems. Those common fields are not public, so this page limits conclusions to the documented BMW and logistics records.

The comparison uses only fields that can be traced to Figure or Agility records and keeps missing values visible. Company-reported hours, parts, tote counts and accuracy remain attributed to the publisher.

The two systems have been reported on different workflows, so cycle time, dexterity and throughput should only be compared after a shared task and measurement protocol are available.

The robots solve different tasks, so hand dexterity and tote throughput should not be converted into one performance score. Public intervention rates, reset policies and comparable full-shift failure data remain missing.

Generation labels also matter. Figure 02 BMW production totals do not establish Figure 03 reliability, and Digit 4 operating hours do not become Digit 5 field hours.

Digit 4 has a longer public record for tote-handling logistics, while Figure has published automotive part handling and a Figure 03 sequencing workflow at BMW. Buyers should match the evidence to the intended task rather than treating either record as proof across all warehouse or factory work.

A procurement comparison should request the delivered robot generation, task definition, shift length, intervention rate, maintenance plan, charging strategy, integration scope and service terms.

Open questions include Figure 03 intervention and uptime data, Digit intervention rate over a full shift, and comparable integration and service costs. Those questions require matched task definitions and reporting that includes failed attempts and resets.

Future deployment reports would be more useful with robot generation, unit count, shift duration, throughput definition, interventions, reset rules and maintenance time published together.