Vibration-based Vehicle Dead Reckoning (VDR) for Localization of Wheeled Vehicles
Masakatsu Kourogi and Takeshi Kurata, National Institute of Advanced Industrial Science and Technology (AIST), Japan
Location: Big Sur
Alternate Number 2
Smartphones become increasingly common for tens of millions of people. Since being equipped with motion sensors, the Global Navigation Satellite System (GNSS), wireless communication devices and high-performance computers, they are highly potential platforms held by the users almost all the time for localization. We have many applications to track and localize the wheel-based vehicles where the user with the smartphone gets aboard.
Although the GNSS is the dominant technology to guide and navigate the wheeled-based vehicles (such as forklifts, cars, buses and railway trains), it is often inadequate or disabled while working and moving in indoor or underground environments.
The Dead Reckoning (DR) is the one of most promising technologies to provide localization solutions while the GNSS cannot be stably used. The DR has been realized by (a) the inertial-based measurements or (b) the rotary encoding systems generating the electrical pulses from the wheels of the vehicles.
However, (a) the inertial-based measurements are still difficult to be achieved even by the today’s high-end smartphones for the vehicle DR.
Although (b) the rotary encoding systems provides stable estimation of velocity of the vehicles, it is difficult to realize DR system from the pulse-counting methodologies since there are no standard protocols to access such information from the wheel-based vehicles (especially from the public transports) without having to tether to the vehicles.
Our research objective is to provide the dead-reckoning for localization of wheel-based vehicles without modifying the wheels, adding special devices or tethering to the wheel-based rotary encoding systems.
Methodologies and Contributions:
Our research uses the finding that the vibration generated by the rotation of wheels shows repetitive patterns whose magnitude has linear correlation with speed of the vehicles. We learned that this phenomenon can be seen in many wheel-based vehicles (such as cars, forklifts, railways, wagons and shopping carts).
We have 3 contributions as below in the methodologies we propose:
1. The DR method by analyzing and measuring the patterns of the vibration generated from the rotary wheels to determine the speed of the vehicle.
2. The DR method to determine the moving direction (namely, forward or backward) by short-term integration of acceleration.
3. The DR method for a smartphone held by the user by analyzing and measuring the attenuated vibration patterns through the human body.
We implement the smartphone-based Vibration-based DR (or VDR) and find that the error rate of the VDR is less than 1% of distance travelled by testing the 3 cars, 15 forklifts, 4 railways and 5 wagons if the smartphone is directly attached to the vehicles. We also find that the error rate is less than 3% of distance travelled if smartphone is held by the user on the vehicle.
Conclusion and Significance of the work:
We learned that the proposed VDR could realize seamless indoor/outdoor localization of wheel-based vehicles in adequate accuracy by the smartphone-grade sensing platform, even if it is held by the user.