What do Pedestrian Dead Reckoning and a Cakewalk have in common?

Pedestrian dead reckoning, while not a ‘cakewalk,’ shouldn’t expect users to walk like balancing cake


Indoor navigation using pedestrian dead reckoning has received much academic and commercial interest over the years. To such an extent that if that is all that people read on the subject, they would conclude that pedestrian dead reckoning has been reduced to “walking while balancing a piece of cake.” That is, the user is expected to keep his mobile device stationary with respect to his body at all times. But let’s face it; we don’t walk like we are balancing a piece of cake.

Pedestrian dead reckoning on smartphones has to be independent of how the phone is carried. The user could be talking on the phone; walking while holding the phone in her hand; keeping her phone in a pocket or a purse; or transitioning between positions. In fact, the user could go through several of these scenarios in the space of a few minutes. She could start with the phone in her purse, then answer a brief phone call, and then keep the phone in her hand. Indoor navigation has to work through all the carrying locations and the associated transitions.

A pedestrian dead reckoning algorithm designed for smartphones has to be able to differentiate between the user’s body travel and her hand movement. In other words, it has to know the difference between a user sweeping her hand to the right versus her making a right turn. We call the direction of travel of the user’s body the “user’s bearing.”

This is one of the key features of our FreeMotion™ Library, the ability to recognize  a CARRY context. Our context awareness algorithm informs our bearing detector how the user is carrying the phone. CARRY can, for example, tell if the phone is in a pocket/container, or held in front of the user, or held at the user’s side, or transitioning from one position to another.

This knowledge allows us to optimize the bearing detection algorithms we are developing that are based on how the phone is carried. Combining user bearing with gait estimation will take us a step closer to the  implementation of an indoor navigation app useful to smartphone users.