Service robots are having enormous variability with their tasks. One thing is common with all of them: they are present in the vicinity and interact with humans. This presence and co-operation capability dictates the design of the robot arm, which is very different compared to the industrial robot designs.
The robot arm should be very light-weight so that inertia while moving will be minimized and the possible impact in case of collision is not harmful. Contradictory to this, the joint links should be as stiff as possible to meet the precision needs of the movements. If the robot arm touches humans, that should be detected by sensors. The possibility of having a configuration in the robot arm so that the arm could lock or even crush a human limb or garment with the specific joint configuration (normally between links 3 and 4) should be prevented by hardware design or at least by software limitations. The robot’s physical design should have no sharp edges in every joint. All joint corners are to be designed with smooth radiuses. All power and signal cables and tubes must be covered inside the arm so that no-one can be stuck to the cable. When making the robot assembly operations, the fixing elements should also be within the covers for the same reason.
These peculiar and contradictory design limitations, together with low hardware cost targets, have resulted in service robot designs looking quite similar. The link configurations are connected through round tubes, all control PCBs and wiring covered inside the joints, motors covered and almost nothing extending from the robot body. The covers are made of plastics and reinforced fibers, to match the light-weight goal.
Nowadays service robots are designed to be quite modular, so they can fit in different work environments. This is why the links also have a modular design. The joint torque and tube length can vary significantly. The configuration can be simulated easily with OSRF’s Gazebo simulation package, which gives the right robot arm configuration for the targeted application very easily. The software, as well as the sensor connections, can also be modular.
The links are at their simplest round tubes, to be most cost effective and not that easy to be gripped. The covers of the joints are of double curved complex shapes. They will most naturally be manufactured by 3D printing. This technology will combine light weight, seat for electronics, cover of the motor, and possibly also a fixing method.
If we look at the statistics regarding the areas where quality problems may occur, the electrical connectors are the most vulnerable in the long run. That is the reason why cables and pneumatic tubes are as long as possible and extra connectors are avoided. This brings in the feeding problem of flexible cables and pneumatic tubes inside the tubular links of the robot arm. The cables should also have extra length to match the needs of the moving robot arm. That extra length must also be handled in a predetermined form, and that is why the cables are twisted. This hassle with cable protrusion is the most complicated and common assemblability problem in service robotics. You want to avoid extra connectors and make the cables as long as possible, but inside the tubes the flexibility of the cables makes the job rather difficult. Also, at the joints the cables must be positioned in a fixed position.
Round tubes have a significant disadvantage. The round shape does not position the joints automatically and accurately to the joints on rotational axis. This needs extra attention and care. Either there must be sharp position marks or some manufactured position seats in the parts. The most difficult procedure is a calibration measurement to have the links aligned accurately.
The design of the final cover fixing elements is the second most common problem. If you apply standard fixing elements, as screws or nuts, their heads will easily pop up from the surface. Those extra features should be avoided so that the uppermost surface is flat and smooth. They are normally covered by a separate cover plate, or there is some kind of assembly construction without any fixing elements. One of this kind is the locking bayonet, being its negative feature a quite difficult manufacturing process.
Car designs nowadays look all the same when competition is extremely harsh. The same seems to be the case for service robotics hardware. The hardware corresponds mainly to the price level, although the real differences are realized by software.
Baya V. and Wood L.: PWC technology forcast, Issue 2, 2015: Future of robots.
Lempiäinen J. et alli: “Integration of Design and Assembly using Augmented reality”, in the book Micro Assembly Technologies and Applications, Springer 2008.