Humanoid Robots: What Lies Behind the Hype Around Optimus, Figure and Others

Humanoid robots in 2024 are development projects with impressive demonstrations but no proven economics in regular operation. Observing them is worthwhile for a company. Investing is worthwhile only when a manufacturer shows sustained output and safety approval in a comparable operation, along with the cost per hour of use. Until then, proven systems such as industrial robots, cobots and autonomous mobile robots decide automation projects.

Videos of Tesla Optimus and Figure as well as Agility Robotics draw millions of views, and board meetings now ask whether the company is missing out. The pressure is understandable. This article puts the systems in context: what they can do and what manufacturers announce, a worked calculation of what a humanoid robot would have to achieve, and a checklist for the decision.

What can humanoid robots do today?

Humanoid robots can walk and pick up light loads, but the demonstrations take place under controlled conditions and no independent measurements from continuous operation exist. Tesla presented the first Optimus prototype at its AI Day in September 2022 and named a target price below 20,000 US dollars at high volume. In December 2023 the company showed a second generation in a video, which according to the manufacturer walks about 30 per cent faster and has hands with tactile sensors in the fingers. Figure, founded in 2022, presented its own prototype in 2023, and Agility Robotics is developing Digit, a two-legged robot for warehouses and logistics that moves totes. All the figures cited come from the manufacturers themselves.

What is actually in use in factories and warehouses is something else. The International Federation of Robotics reported in September 2023 a record 553,052 newly installed industrial robots worldwide in 2022, five per cent more than the year before. Around 73 per cent of them went to Asia. These machines have no legs and no hands, and they have worked reliably for decades. The article on robotics in mid-sized companies shows where robots already pay off.

Why the human form, and what speaks against it?

The human form has one real advantage: a humanoid robot can work in environments built for people, with stairs and doors as well as shelves reached by hands. That saves rebuilding a hall. Against this stand drawbacks that every operator knows. Two legs need active balance and therefore energy that a wheel or a fixed base does not consume. Hands with many joints are complex and prone to failure. Battery runtime is limited, and a falling robot of human size is a safety hazard.

Standards set limits as well. The safety rules for industrial robots (ISO 10218) and for collaboration between humans and robots (ISO/TS 15066) were written for robot arms at fixed stations. A dynamically balancing biped working beside employees raises questions these rules do not answer, and no robot runs in regular operation without approval from the safety specialists. How robots can work safely beside people is described in the article on cobots.

When would a humanoid robot pay back?

A humanoid robot pays back when the annual saving from replaced working hours exceeds running costs and the investment is recovered within a period that fits its service life. The calculation below is an example with assumed values and not a market forecast. It shows which variables determine the result.

The example shows where the decision falls. The price of the robot influences the result less than the performance factor and the running costs. A device that works at half the speed of a person and needs constant supervision does not pay back over its service life, even if it is cheap. For manufacturers this is the real hurdle, and it explains why the videos rarely show timings for complete work steps.

Comparing robot types for your operation

Which type fits a task depends on the environment, the maturity and the approval of the safety specialists. The table sorts the four types by these criteria.

Anyone looking to automate a warehouse will find a calculation for the type with proven economics in the article on automated guided vehicles.

How should executives approach the topic?

Executives should observe the topic and set fixed trigger criteria for doing so. As in every technology cycle, announcements are declarations of intent, and only what runs in a third party's operation is proven. The article on the AI hype cycle describes the features that separate progress from announcement.

If all five points are met, a limited pilot at one workstation with a clear metric is worthwhile. Until then, observation costs only a few hours a quarter, in which one responsible person evaluates manufacturer statements and operator reports.

Humanoid robots and Niehues Consulting

Niehues Consulting is a consultancy for AI management and organisation design, seen from the delivery perspective, based in Düsseldorf. Founder Marin Niehues brings new technology into operation, in projects ranging from VC-funded scale-ups to DAX 40 corporations and in more than 20 countries on six continents. Niehues Consulting does not supply robots or integrate cells. It supports business case and introduction as well as delivery management and organisation, which means the question of whether and when a purchase pays back and how the operation absorbs it.

The approach follows the Niehues Delivery Model. For robotics initiatives, three of its test questions matter most: which business result should change measurably? Does the prototype prove the benefit in the real process with real data? Who is responsible for operations from day one? The starting point is a free initial consultation of 30 minutes, which you can arrange via the contact page. Scope and terms of an engagement are discussed there.