How to “Train” Your Sensor

It is becoming much more common and socially acceptable to wear smart devices like fitness trackers and smartwatches that come loaded with sensors and compute resources to assist us in myriad ways in our daily lives. Miniaturization of electronic components is accelerating this trend by making devices more comfortable to wear, less expensive, and more feature-rich. But while a single wearable may be capable of doing everything that a user wants it to do, different applications require that the device be located in a different position on the wearer’s body. The collar, for example, is a good location for voice commands. To monitor breathing, a sensor would ideally be located on the front or back of the upper body. But to sense hand inputs, the forearm or hand is a better choice.

So while a single device could handle all of these use cases, we still need multiple wearables to accommodate the necessary positioning of instrumentation. An early prototype device that may one day lead to a better solution to this problem has recently been described by a team of researchers at the University of Maryland, College Park and the University of California, San Diego. Their system, called Calico, uses a flexible on-cloth track that allows a smallish robot to crawl along the body to a precise location on demand. Tracks can even be magnetically connected to other articles of clothing (e.g.: linking a shirt with pants) to extend the range of the robot.

Calico’s on-cloth track system is made up of custom soft tracks stitched into clothing. These give the robot a consistent, easily traversable path regardless of the type of material the clothing is made of, or natural deformations that are present in that material. The system also has rotating switches, reminiscent of railway turntables, that can turn the robot in any direction to allow it to select among different available paths to take. This makes every part of the body accessible, however, since adding more tracks alters the appearance and weight of the clothing, tracks must be carefully laid out to minimize the total length used.

The robot has a four-wheel design with two Pololu Micro Metal Gearmotors to provide the power for locomotion. A wheel-track pinching mechanism is employed to keep the robot on the tracks and allow it to climb against gravity. To control the robot and add sensing, a custom PCB was designed with a MDBT42Q microcontroller, which also adds support for Bluetooth Low Energy wireless communications. An inertial measurement unit was included in the design to give the robot some basic sensing capabilities. A pair of hall-effect sensors on the bottom of the robot help it to sense where it is on the track, and also serve to control the track’s rotating switches.

To showcase Calico’s capabilities, the team set up a few scenarios to demonstrate the robot in action. In one case, an electret microphone with a stethoscope head was added on to the robot. By positioning itself correctly, this device could collect on-demand chest auscultations, which in theory could be transmitted to healthcare professionals to keep tabs on a patient’s health without a visit to the clinic. In another demonstration, the researcher created a smartphone app that pairs with the robot to serve as a workout tracker. By relocating to an appropriate part of the body (e.g.: knee for squats, elbow for push ups), the tracker can monitor the performance of an exercise with a high degree of precision.

Calico unlocks some very interesting possibilities for the future of wearable devices, but in its present form, it does not seem entirely practical. Unless you want to look like Chewbacca with his bandolier strap — oh, and a rather large robot driving around on it — you will need to wait for this technology to be made much more compact and unobtrusive. Then again, Halloween is right around the corner.