Evaluation of humanoid robots for surface finishing

Sanding, polishing, sandblasting, sanding, coating and painting - these manufacturing processes are used in a wide variety of industries. From an ergonomic point of view, such tasks are a real challenge and pose serious risks to human health.

It is not surprising that companies are increasingly thinking about automating these operations with the help of robots. When the parts are large, the manipulator has to be constantly rearranged in order to process the entire surface.

People often ask: wouldn’t humanoid robots be suitable for working with large parts? Let's look at their features and see how applicable they are in production processes.

Why legs are not needed in the workshop

Feet are not the most efficient way to move on the flat floor of a factory floor. In addition, the robot often has to be tethered with a cable to power a tool (say, a grinding disc). If a walking platform needs such a “leash,” you can forget about flexibility.

It is much more convenient to use fixed rails on the floor or a mobile base - they are better suited for transporting manipulators when processing parts. Legs only complicate the design and increase safety risks without providing any benefits. Conclusion: On a factory floor, feet are not needed for production operations.

Multi-fingered hands: unnecessary complexity

Multi-fingered hands provide flexibility and dexterity when manipulating objects. But they are expensive. And many production processes - grinding, sanding, sandblasting, coating, tightening screws - come down to working with tools.

Holding, say, a sander with a multi-fingered brush is not a good idea: the grip may not be strong enough to work at high speed. Therefore, it makes more sense to attach the tool directly to the manipulator through a simple adapter - this way we reduce the cost by getting rid of complex brushes.

Head: Useless for production

The head of the humanoid robot is small, so the cameras in it have to be placed close to each other. To ensure good visibility and accuracy, it is best to distribute cameras (and other sensors) throughout the work area.

Often a sensor mounted on a manipulator next to the tool is much more effective. In addition, cameras can simply be placed above the part in the workshop. So heads are not particularly useful in production.

Two hands: correct configuration

But two hands are a useful thing when processing large parts. To increase coverage and speed up work, manipulators need to be placed far enough apart to reduce the risk of collisions and maximize work space. This is not the same configuration as humanoid robots.

The optimal scheme for surface finishing is shown in the figure below. It uses two hands. There is a mobile solution for rearranging them. There is vision to understand the scene. But the configuration of components here is not at all the same as in humanoid robots. Many people argue that it is androids that provide economies of scale.

The diagram shown above contains common industrial manipulators, rails, cameras and force sensors. And most of the hardware in such a cell benefits from economies of scale. So humanoid robots are not at all necessary to achieve this effect in production.

Most production tasks focus on one thing: reducing cycle time and increasing throughput. Therefore, universal industrial robots that do not care about human limitations are a very attractive option here.

Such robots work 4-5 times faster than humans and can exert much greater effort. They have already proven their reliability in the harshest conditions. So assembling production cells by arranging industrial manipulators in the desired configuration seems to be a winning solution for most problems. This means that androids are unlikely to become winners where the main criteria are accuracy, performance and speed.