The unique design feature of the actuator that I designed is it's use of a planetary gearbox as a source of transmission. Many of the exoskeleton actuators that are being used in both commercial and research applications at this time are using harmonic drives as a transmission. Harmonic drives are a very special type of transmission, designed specifically for high precision robotic applications.
Harmonic drives do have their drawbacks as well though. Because the two splines are in constant rolling contact with one another and have to be strained to do so, harmonic drives are very high friction systems. This means it takes a considerable force to start them rotating, and once up to speed, a considerable amount of the energy put into the system is lost due to friction. This is less of an issue in applications where harmonic drives are typically used such as robotic arms in industrial settings because they are connected to the power grid and energy usage and weight are less important than size and precision. Because the exoskeleton is a mobile robot, these priorities shift in favor of a more efficient system that can more efficiently convert the electrical power in the batteries into mechanical power. This is where the planetary gearbox comes in. Planetary gearboxes are significantly higher mechanical efficiency systems than harmonic drives that can more efficiently use the stored energy at the cost of a small drop in precision due to backlash of the gears.
From there the motor was selected. The main constraints on the motor in this case are the voltage, current, and size of the motor. The output voltage and peak current supply of the controls system had already been set at this point in the project and needed to be designed around. The motor would need to be as flat as possible while still providing enough torque. This is so the actuator doesn't stick out too far from the user during use. A hubless, brushless, DC motor was selected that fit the requirements with 1 Nm peak of torque.
With the motor settled, the gear ratio needed to be determined. In order to get 50 Nm peak out of the actuator, the transmission had to be 50:1. This actually isn't as simple as it sounds due to the nature of planetary gear systems. I will leave out most of the calculations in order to keep this section brief, but more information can be found here. I will explain that it is very difficult to achieve a single stage planetary reduction of more than 10:1, so I knew this would have to be a multi-stage transmission. I settled on using a set of spur gears to achieve an initial 2:1 reduction and two 5:1 planetary stages for a total of 50:1
In order to ensure the gear teeth could handle the expected load of the actuator, the Lewis gear strength equation was used. Again, I don't want to go into too much detail on this, but the idea behind the equation is to estimate the stress on the gear teeth by treating them as cantilever beams with a static load on the end. The stress on each tooth is based on the pitch of the gear, the width of the gear face, the number of teeth in the gear, and the load on the tooth. As the teeth get smaller or the load increases, so does the stress on the gear teeth.
Finally, the efficiency needed to be calculated. Like I mentioned in the intro, the main consideration driving the use of a planetary system over a comparable harmonic drive is the efficiency. This was done based off of the work of a few scientific papers that I found that used the known mechanical losses of spur gears and transformed them into overall mechanical efficiencies of a planetary set up. I don't have a good link for this one that is publicly accessible as most were higher level research papers, but the idea is that there has been a lot of research into the efficiency of various sizes of mating spur gears. By treating the two sets of meshing gears (sun to planet, and planet to ring) as two sets of spur gears you can generate a basic efficiency. Based off of which gear is your input and which is your output, you can estimate the overall efficiency of the stage. This efficiency calculation was done for the spur gear stage and the two planetary stages to combine for a theoretical efficiency of around 95%.
The housings are designed in a way to ensure that all of the parts align concentrically to one another. The two housings each have a lap joint around the perimeter to align the housings to eachother (see below). All of the key features of the housings that align the components are designed so that they can be machined in a single operation on a CNC mill. This setup makes it significantly easier to ensure that all of the parts align with one another to prevent binding issues.
Manufacturing and Assembly
Testing and Validation
The speed vs current test consisted of slowly increasing the current applied to the motor while monitoring the motor speed using the built in Hall sensors. This was transformed into an output speed of the actuator using the gear ratio. The output arm was removed to allow the motor to spin freely without hitting the hard stops on the housing an completely unload the actuator. This test was preformed on the actuator I made as well as a comparable harmonic drive based actuator using the same motor and a 50:1 harmonic drive as a transmission. The results of the test can be seen below.
The other test that was performed on the actuator was a measurement of the output torque of the actuator versus input current to the motor. This was done by applying the current in the same fashion as the first test and monitor the output torque using an external load cell. The current was stepped at 1A increments and the torque measured and plotted below.
This result is vital for the future of planetary style transmissions in powered orthotic applications. The benefits of high efficiency and ease of back-drive-ability more than compensate for the slight reduction in output precision due to backlash and time invested into designing a custom system such as this one. Although I have finished this project, its impacts will live on in the lab that I worked in and hopefully influence the future course of powered exoskeletons.