Mercedes-Benz confirms that Apollo robots collected data in a real production environment and that employees taught them production tasks using teleoperation and augmented reality before further autonomy training.
HU-R010 / Deployment watch
Apollo × Mercedes-Benz
An operator-confirmed production-environment training program with an unusually clear teleoperation-to-autonomy boundary, but no published operating or reliability results.
What this means
The short version, without reliability jargon
The sources do not say how many robots were used, how long they operated, how many tasks they completed, how much work was autonomous or how often people had to help or recover them.
This is meaningful evidence of real factory integration and a documented path from human teaching toward autonomy. It is not yet evidence that Apollo can perform the work independently or reliably over time.
Evidence verdict
Established. Claimed. Unknown.
- Project and siteApollo training at the Mercedes-Benz Digital Factory Campus in Berlin-Marienfelde[2]
- Intended tasksComponent transport, line delivery and initial component quality checks[2] [1]
- Production-environment dataApollo robots collected data in a production environment for specific MO360 use cases[2]
- Human teaching modeMercedes-Benz employees transferred production knowledge through teleoperation and augmented reality[2]
- Autonomy stageMercedes-Benz says the robots are learning to act autonomously at the campus[2]
- Bounded observation window and operating hours
- Autonomous and teleoperated time shares, supervision and intervention count
- Task attempts, output, success criteria and production throughput
- Exact fleet and hardware versions, downtime, failures and recovery
What the evidence says
Mercedes-Benz confirms that Apollo robots collected data in a real production-development environment at its Berlin-Marienfelde Digital Factory Campus. Employees taught production knowledge through teleoperation and augmented reality, and Mercedes says the robots are learning to act autonomously. A later Apptronik update names Mercedes-Benz within broader mixed teleoperation and autonomous data-collection workflows. None of the reviewed sources publishes operating hours, completed tasks, fleet size, autonomy share, intervention frequency, downtime, failures, recovery time or availability.
01 / Reliability fields
What is known—and missing
02 / Claims and evidence
What each source supports
Apollo training at the Mercedes-Benz Digital Factory Campus in Berlin-Marienfelde
Mercedes-Benz identifies the robot, operator and production-development site. This establishes a real production environment, not routine autonomous production or reliability performance.
Sources: [2]Component transport, line delivery and initial component quality checks
Mercedes-Benz and Apptronik describe closely related intralogistics and inspection use cases. Neither source publishes completed-task counts, attempts, throughput, error criteria or production output.
Sources: [2] [1]Apollo robots collected data in a production environment for specific MO360 use cases
The operator confirms physical data collection in the factory-development environment. It gives no operating-hour total, observation window, robot count, task denominator or indication that the collected time was uninterrupted.
Sources: [2]Mercedes-Benz employees transferred production knowledge through TeleoperationA person controls some or all robot actions from a distance rather than the robot selecting those actions by itself.Here, teleoperation describes how Mercedes-Benz employees demonstrated production tasks during training. It is not a published count of interventions during autonomous work.Learn more → and augmented reality
Teleoperation is explicitly part of task teaching. The source does not quantify teleoperated time, supervision, interventions during autonomous execution or how often a person had to recover the robot.
Sources: [2]Mercedes-Benz says the robots are learning to act autonomously at the campus
This is evidence of an autonomy-development stage, not proof of unsupervised production work. No autonomous-time share, task-success rate or transition date is disclosed.
Sources: [2]Apptronik names Mercedes-Benz among customer sites using data-collection workflows with teleoperation and autonomous execution
The update describes a broader Apollo 2 data-collection network and does not isolate the Mercedes site, task, robot count, operating time or share of each operating mode. HumanoidUptime does not apply the Apollo 2 configuration retroactively to the earlier Mercedes activity.
Sources: [3]Apptronik announced a commercial agreement to explore Apollo manufacturing applications in March 2024
The agreement dates the public program and planned use cases. It does not disclose a purchase quantity, completed deployment, paid production service or reliability result.
Sources: [1]03 / Timeline
What changed, and when
-
Apptronik announces the Mercedes-Benz agreement
The manufacturer describes planned component delivery, line-side logistics and inspection applications without publishing a fleet commitment or operating result.
[1] -
Mercedes-Benz documents production-environment training
The operator says Apollo robots collected factory data, employees transferred task knowledge through teleoperation and augmented reality, and the robots were learning autonomous action at the Digital Factory Campus.
[2] -
Apptronik describes broader mixed-mode data collection
Apptronik lists Mercedes-Benz among customer and partner sites using related data-collection workflows with teleoperation and autonomous execution, but reports no Mercedes-specific performance result.
[3]
04 / Sources
Trace every finding
-
Source 1Manufacturer statement
Apptronik and Mercedes-Benz enter commercial agreement ↗
Apptronik · Published 2024-03-15 · Accessed 2026-09-04
-
Source 2Operator statement
AI and humanoid robots at the Digital Factory Campus ↗
Mercedes-Benz Group · Published 2025-03-18 · Accessed 2026-09-04
-
Source 3Manufacturer statement
Welcome to Robot Park, where Apptronik's Apollo goes to work ↗
Apptronik · Published 2026-06-30 · Accessed 2026-09-04