Information of road conditions such as gaps, angle of inclination and gravel types is very useful for realizing convenient pedestrian navigation systems because some users can not walk on a path with bumps and steep slopes. In this research, I design and develop a road condition recognition algorithm using shoe-mounted inertial sensors. The proposed method estimates angle of inclination (AoI), stability and smoothness of the road. Since feet are the most reflective part of road conditions, the proposed method uses a pair of small sensor boxes mounted on feet with accelerometers and gyro sensors inside. At first, the proposed method detects stationary stance phases (foot-flat phase) in which the foot clearly reflects road conditions. Since it is assumed that feet do not move in stance phases, the stance phases can be detected by accelerometers and gyro sensors attached to the shoes. Then, based on this information, the method estimates AoI based on a distribution of gravity on each axis of accelerometer. Moreover, since an uneven road (e.g. stone-paved) causes fluctuation of AoI, road smoothness can be estimated based on variances of AoIs from multiple steps. Meanwhile, feet do not stop perfectly during walking on unstable road (e.g. gravel or mud) because feet move according to collapsing of the ground. Based on this fact, the proposed method can estimate road stability using variance of acceleration.
List of Publications
- Takumi Satoh，Akihito Hiromori，Hirozumi Yamaguchi，Teruo Higashino : " A Novel Estimation Method of Road Condition for Pedestrian Navigation " , International Workshop on the Impact of Human Mobility in Pervasive Systems and Applications (PerMoby '15) ， Mar, 2015 ．
Core Body Temperature Estimation
Monitoring core body temperature is important to evaluate the effect of heat environment and human activity on human body such as heat stroke prevention and athlete performance evaluation. The core temperature is often measured as rectal temperature or tympanic temperature, but both of them are not usually acceptable in daily life due to their invasiveness.
In this research, we propose a method to estimate core temperature during exercise based on the human thermal model considering individual differences using wearable sensors. The method employs the two-node model, which is a human thermal model to simulate the change of core and skin temperature by calculating heat exchange between core, skin and environment and heat production in the body. As the model involves several parameters to consider the difference of perspiration rates and skin blood flows in response to heat between individuals, finding the appropriate values for each individual is essential for accurate estimation of core temperature.
The method determines the optimal parameter set representing individual thermoregulation function and physical feature through real-time monitoring of skin temperature, ambient temperature and humidity, and heart rate as activity intensity. The measured values except the skin temperature are input to the two-node model and skin temperature is used as reference of calibration. Based on these input, we conduct thermal simulations and obtain calculated skin temperature and core temperature. The temperature as a result of simulation depends on a set of parameters. Then, we conducted exhaustive simulation for all the possible sets of individual parameters and obtain the optimal parameter set which minimizes the squared error between simulated and measured skin temperature.