Step-by-step dead reckoner

Imagine you walk in front of a painting in an art museum, and your smart-phone knows you are facing that particular painting and gives you the opportunity to get an explanation of the painting (see Figure 1).  Our prototype product provides the necessary location and heading awareness to enable such applications.

We have connected, via a Bluetooth module, a smart-phone to a coin-size 9-axis enhanced Inertial Measurement Unit (IMU).  The IMU unit must be attached to the user's shoe (see Figures 2 and 3).  It includes 3 axes of accelerometers, 3 axes of gyroscopic sensors, and also 3 axes of magnetometers.

Processed magnetometer data provides a heading reference and corrects drift in the gyroscopic sensors.  The gyroscopic sensors allow fine-grained orientation data to be computed.  Given the orientation data, the accelerometers allow precise step lengths to be computed.  Between steps, the shoe-mounted IMU is stationary for a period of time.  This allows linear (and angular) velocity to be reset to zero after each step, significantly reducing the drift error inherent in MEMS-based accelerometers (and gyroscopic sensors).  Knowing the direction and length of each step, we compute the user's present location from a known starting point.  The starting point might be the user's last known GPS location, for example when the user entered the art museum from the outside.  Others have demonstrated that such techniques, on their own, can measure distance traveled with an error of less than 1%.  We believe that we can improve the algorithms to significantly reduce the error.  Our prototype also determines the direction the user's foot is pointing. Therefore in a museum, for example, the user could select one from a set of neighbouring paintings by simply pointing his or her foot towards the painting.  Our location awareness prototype should nicely complement other positioning technology, such as the Nokia Beta Labs project [1] at the Kamppi Shopping Centre that determines rough location from the WiFi access points detected around the user.  Our product senses accurate fine-grained relative location, and technology such as the Kamppi project can help to correct any cumulative errors that could build up using our product.  In addition, the floor plan constrains the paths people can take through a building and therefore floor plan information can further help to correct any accumulating error.  Ideally, software similar to GeoClue [2] can link our product with these other location technologies.  Note that our product on its own, and even when combined with floorplan/map correlation software, works without any required infrastructure, such as beacons.

Because of the prototype nature of our hardware, we have attached the IMU to the side of the sole of the user's shoe.  However, the IMU is actually small enough to be embedded in the sole of a shoe, just like the way the Nike+ sensor [3] can be embedded in some Nike shoes.  Assuming the costs of MEMS-based IMUs continue to drop, our system could soon be affordable as a consumer product.

While the IMU data is being received by the smart-phone via the Bluetooth module, our C/C++ backend software performs the necessary calculations to determine heading and relative location of the user.  We have also written a demo application to display the user’s current location and information about objects the user chooses to face.

We gratefully acknowledge the support of Nokia Products Limited.

 

Media

Figure 1: Artist's sketch, showing a museum patron using her smart-phone to view contextual information about the exhibit she is near, with her location and heading being provided by a concept device realized by our project



Figure 2: Our prototype hardware and software in use – the software is providing information on a piece of artwork that the user walked in front of.



Figure 3: Our chosen sensor module mounted on a development board sits beside our prototype software running on a Nokia N900 smart phone. The Bluetooth module and battery is out of the frame of the photograph.

 

Contact

Dr. Craig Scratchley, P.Eng. -- wcs@sfu.ca

This page hosted by Simon Fraser University.