Inspecting the condition of dykes and other sea defense structures is typically a task for a team of robots. They consume a lot of energy to move across the dykes, perform tests, and communicate the results for six hours a day. Because charging stations are not a realistic scenario, University of Twente researcher Douwe Dresscher looked at making the robot as energy autonomous as possible. He obtained good results by having the robot store mechanical — rather than electrical — energy, and by introducing an innovative automatic gear box. The gear box is a modern version of the “variomatic” model used in Dutch DAF automobiles. While the variomatic uses a belt drive, the inspection robot uses two metal hemispheres.
The first factor Dresscher had to consider was the best way for the robot to move on the dyke — using wheels, caterpillar tracks, or legs. Wheels work well and are energy efficient on an even surface, but a wet and muddy slope is something quite different. Tracks are more powerful in this case, but they can damage the dyke by the way they turn and they are energy inefficient. A walking robot with four to six legs would perform best. Walking robots do consume a lot of energy, and existing commercial walking robots always wear a big battery pack.
Electromotors are primarily responsible for this high level of energy consumption. They perform best at high revolution speeds and low torques, but in the walking movement they often work at low revolutions and high torques instead. By storing energy in a me chanical rather than an electrical way, the electromotors can do their job using the best operation regimen, and mechanical energy can be reused. This is what Dresscher calls Controlled Passive Actuation.
The system can store mechanical energy in a spring, for instance. A gear box takes care of the optimum transmission. Two half-turning half hemispheres therefore are constantly in contact. The angle changes when the torque changes — resulting in another relative radius. The difference in effective radiuses determines the transfer ratio and the best mechanical load. The electromotors join in only to compensate for mechanical losses. By doing so, they can work within the high-rev, low-torque system.
To eventually make the dyke robots fully self-supporting (sensing and communication also require energy), they will need to “harvest” energy when moving. Solar energy, wind energy and biomass are viable options. Dresscher didn’t examine this aspect in detail; his work focused on the locomotion part of the robot and the powertrain. The new powertrain is specifically designed for future dyke inspection robots, but it could also be applied to improving the energy efficiency of existing robots and robot arms.