Integrating Power in Your Robot Design

On a recent visit to an MIT spinoff that specializes in rehabilitation robotics, I had the opportunity to try on a partial exoskeleton. The compact motors and lightweight, anodized aluminum frame — while not quite up to the standards of the Iron Man costume — were a testament to the artistic, ergonomic, and engineering expertise of the development team. However, when it came time to activate the unit, someone pulled out an encyclopedia-sized NiMh battery pack and massive cable to power the exoskeleton. It reminded me of the svelte laptops that ship with huge power bricks and the miniature, wireless spy cameras that require a relatively huge 9V battery.

You can see the same design approach on YouTube, where Japanese researchers display their latest ultra-thin, unobtrusive, robotic exoskeletons to help elderly navigate stairs or simply walk. While the motors, sensors, and exoskeleton components are impressive, the camera never seems to capture the power source. I assume that a truck battery is tucked away behind a curtain somewhere on the set. In the real world, however, customers do see and deal with battery packs.

As the market has illustrated, if your robot demands so much from a small battery pack that it overheats or ships with a battery that weighs more than your robot, the odds of commercial success are poor.

What’s the point? In retrospect, the MIT spinoff could have saved scarce development time and resources by focusing on the battery power requirements early on in the design process. Why spend a $50 premium on an electric motor that’s 100 grams lighter when that same $50 could reduce battery weight by 250 grams? It’s like the cyclists who pay a premium for an ultra-light bike and then invest in a super heavyweight bike lock to protect it. Better to buy a less expensive, somewhat heavier bike and a regular heavyweight bike lock.

So, how do you go about integrating power requirements in your next robot design project, as opposed to searching for a solution as an afterthought? My approach is to set up a spreadsheet listing the cost, weight, and power requirements for every component going into my robot design. Whereas sensors and actuators have explicit power requirements, many components don’t. In these cases, weight is a good relative indicator of power requirements.

For example, given two control arms — one 400 grams and another 800 grams — you can assume that the heavier arm will require more power to move. How much more depends on where it’s used in the robot and whether the robot is stationary or moves about on legs or wheels.

I also set up a table that lists battery packs and specifications including cost, capacity, weight, and dimensions. I start with a battery pack that seems like a good compromise in specifications and work the dimensions, weight, and capacity of the battery pack into the robot design. In the spreadsheet format, I can interactively perform what-if analyses. Moreover, the battery pack entry in the table is always in clear view, alongside the CPU or microcontroller, sensors, and other components.

This technique isn’t limited to traditional battery packs. In fact, an integrated approach to designing in power from the start is often mandatory in robot designs that feature solar cell, fuel cell, and other non-traditional sources of power. You probably don’t want to tuck a solar cell panel on the underside of a crawler, or use a fuel cell in an airtight robot designed for amphibious environments.

In my experience, the most critical applications of power integration are in the remote control and semi-autonomous aircraft, where the mass, size, and capacity of the battery defines the operating constraints of the aircraft. Regardless of the AI and machine intelligence on board, when the battery pack is exhausted, the aircraft is going down.

Bottom line — think of the battery or other power supply as an integral component of your robot design. One that’s just as important as the onboard processing or sensor arrays.  SV

Posted by Michael Kaudze on 08/21 at 03:52 PM


<< Back to blog