Chandra Tjhai

Mr. Chandra Tjhai

orcid

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 September 2017
Verified
Conference Paper
Closed access
CTChandra Tjhai
Chandra Tjhai
KOKyle O'Keefe
Kyle O'Keefe
This paper presents a sensor fusion algorithm to estimate step-size for pedestrian navigation using multiple low-cost wearable inertial sensors. A sensor data logger is developed for reading multiple low-cost sensors. A treadmill experiment with an optical motion capture system is conducted for algorithm evaluation. A INS-ZUPT Kalman filter is implemented to estimate stride length from each foot mounted IMU. In addition, the pitch angles of the thighs and shanks are used in a step length estimation algorithm. The results of these methods are compared to previously developed algorithms. The results show that the low-cost multiple-sensor algorithm is capable of producing a step-size estimation that is comparable to previously developed methods.

17 July 2019 in Sensors
Verified
Journal Article
Open access
PDF file
CTChandra Tjhai
Chandra Tjhai
KOKyle O'Keefe
Kyle O'Keefe
This paper demonstrates the use of multiple low-cost inertial/magnetic sensors as a pedestrian navigation system for indoor positioning. This research looks at the problem of pedestrian navigation in a practical manner by investigating dead-reckoning methods using low-cost sensors. This work uses the estimated sensor orientation angles to compute the step size from the kinematics of a skeletal model. The orientations of limbs are represented by the tilt angles estimated from the inertial measurements, especially the pitch angle. In addition, different step size estimation methods are compared. A sensor data logging system is developed in order to record all motion data from every limb segment using a single platform and similar types of sensors. A skeletal model of five segments is chosen to model the forward kinematics of the lower limbs. A treadmill walk experiment with an optical motion capture system is conducted for algorithm evaluation. The mean error of the estimated orientation angles of the limbs is less than 6 degrees. The results show that the step length mean error is 3.2 cm, the left stride length mean error is 12.5 cm, and the right stride length mean error is 9 cm. The expected positioning error is less than 5% of the total distance travelled.

1 September 2018
Verified
Conference Paper
Closed access
CTChandra Tjhai
Chandra Tjhai
KOKyle O'Keefe
Kyle O'Keefe
This paper presents heading estimations from multiple low-cost wearable sensors distributed on the lower limb segments. A low-cost commercial motion capture suit from Enflux is used to record accelerometer, gyroscope and magnetometer measurements. Roll and pitch angles from each sensor are estimated to level each magnetometer. The sensor orientations are computed using a Kalman filter. The step length is computed using the sensor mounted on the pelvis while the stride length is computed using the foot-mounted sensors. The results show that the pelvis is the best location to track pedestrian heading while other sensors have poor performance due to difficulty in estimating the roll and pitch angles.

1 April 2018
Verified
Conference Paper
Closed access
CTChandra Tjhai
Chandra Tjhai
JSJeremy Steward
Jeremy Steward
DLDerek Lichti
Derek Lichti
KOKyle O'Keefe
Kyle O'Keefe
This paper presents a comparison of joint angle and step length estimation computed from multiple low-cost wearable sensors and optical systems. Seven wearable sensors are mounted on the pelvis, and the thigh, shank, and foot of both legs. Two types of experiments are conducted, a treadmill walk and straight line walk. The joint angles are estimated using a Kalman filtering method to fuse all sensor measurements. The step lengths are computed based on the estimated joint angles. The results show that the average errors of pitch angles and step lengths are 3 degrees and 6 cm, respectively, for treadmill walk.

1 September 2018
Verified
Conference Paper
Closed access
SNSharareh Naghdi
Sharareh Naghdi
CTChandra Tjhai
Chandra Tjhai
KOKyle O'KEefe
Kyle O'KEefe
The utility of a commercially available low cost ultra-wideband (UWB) real-time location system (RTLS) to provide reference position information for the generation of accuracy high-density radio maps is investigated. The UWB RTLS hardware is deployed in a medium-sized office space in a university building and UWB ranges are recorded in static and kinematic mode over an array of precisely surveyed locations within the room. WLAN received signal strength indicator (RSSI) observations are taken simultaneously to UWB range measurements. Radio maps are then generated by interpolating the observed RSSI measurements using both the precisely surveyed coordinates and the UWB RTLS coordinates. The maps are compared and the feasibility of the method is discussed.

14 February 2022
Verified
Conference Paper
Closed access
PDF file
PXPatrick Xu
Patrick Xu
CHCurtis Hay
Curtis Hay
RKRakesh Kumar
Rakesh Kumar
ISIqbal Surti
Iqbal Surti
CTChandra Tjhai
Chandra Tjhai
Most Level 2 autonomous driving systems including General Motors (GM)’ Advanced Driver Assistance System (ADAS) employ multiple sensors to achieve lane-level localization. As the only source onboard that provides absolute 3D global poses, Global Navigation Satellite Systems (GNSS) is susceptible to various Radio Frequency (RF) interferences such as jamming, spoofing and meaconing, intentionally or unintentionally. Since these interferences pose a growing threat to the safety and reliability of the automated vehicles, detecting, characterizing, and mitigating them become increasingly important. This paper proposes a machine learning-based method to detect spoofing signals. Basically, it treats the detection as a binary classification and evaluates the potential of applying supervised machine learning in the process. Various classifiers are trained and evaluated, and the best performing model is further optimized to predict the existence of interference signals. Raw GNSS measurements such as Pseudorange, Carrier Phase, Doppler, Clock Bias and Drift, Carrier-to-Noise Density (CN0), as well as their respective measurement uncertainties, are combined to train and test the machine learning models. Data cleaning, feature engineering, variable correlation analysis, and principal component analysis (PCA) were performed before the training process. Finally, to evaluate the effectiveness of the proposed method, two additional rounds of drive tests with real meaconing signals re-transmitted by multiple GNSS repeaters were performed at a GM vehicle testing facility in Michigan, US. Validation results showed an average of 97.4% detection accuracy, as well as an overall F1 score of 0.937.