AUTONOMOUS MOBILITY

Efficient Clothoid Tree-Based Local Path Planning

We propose a local path planning method for self-driving robots using a curvature-continuous clothoid tree. Clothoids are suitable for road design and path planning for wheeled vehicles, as they have linear curvature profiles. However, their real-time application faces challenges due to the lack of a closed-form expression. To overcome this, we introduce two techniques: 1) accurate and smooth approximation of clothoids using Gauss-Legendre quadrature; and 2) data-efficient parameterization of interpolating clothoid splines based on curve symmetry and similarity. These techniques are demonstrated with numerical examples and integrated into our path planning algorithm, improving accuracy and efficiency across various driving scenarios. (ICRA 2024)

Planned Trajectory Classification to Prevent Rollover and Slip

We propose a novel planned trajectory classification method (PTCM) to evaluate the safety of the car-like four-wheeled mobile robots (4-WMRs) with Ackermann steering and high-stiffness suspensions. To classify a planned trajectory to be safe or unsafe before the 4-WMR actually follows it, the conditions of the wheel forces (WFs) necessary to execute the planned trajectory without rollover and slip are calculated using the passive decomposition of the WMR dynamics with the Pfaffian constraints of the no-rollover and no-slip conditions. Similar to the case of Navier's table problem, only nine-dimensional WFs projected onto the constrained space are identifiable among the twelve-dimensional WFs. This indeterminacy turns out not to affect the rollover prediction, yet does so for the slip prediction. For this, we propose novel optimistic and pessimistic methods, together upper and lower bounding the exact slip prediction. The proposed PTCM classifies the planned trajectory as safe if rollover and slip are not predicted and unsafe otherwise. (RA-L 2023)

Legged Robot Neck Design for Robust Visual-Inertial State Estimation

We propose a new neck design for legged robots to achieve robust visual-inertial state estimation in dynamic locomotion. While visual-inertial state estimation is widely used in robotics, it has a problem of being disturbed by the impacts and vibration generated when legged robots move dynamically. To address this problem, we develop a tunable neck system that absorbs the impacts and vibration during diverse gait locomotions. This neck system consists of two components: 1) a suspension mechanism that compensates for the weight of the head equipped with a camera and IMU (inertial measurement unit), absorbs the impacts and the head motion of high frequencies including vibration as a fixed low-pass filter; and 2) a dynamic vibration absorber (DVA) that can be reactively-adjusted to diverse gait frequencies to alleviate excessive head movements. We present a dynamics analysis of the neck system and show how to adjust the target frequency of the system. (RA-L 2023, IEEE Spectrum)

Flying-LASDRA: Autonomous Outdoor Flying

We perform outdoor autonomous flying experiment of flying-LASDRA (aka f-LASDRA) system, constructed with multiple ODAR-8 links (https://youtu.be/S3i9NspWtr0) connected via cable with each other. Each ODAR-8 can compensate for its own weight, rendering f-LASDRA scalable. Utilizing SCKF with IMU/GNSS-module on each link and inter-link kinematic-constraints, we attain estimation accuracy suitable for the stable control of flying while maintaining internal shape by using only standard GNSS, not RTK-GPS (<5cm: cf. 1-5m w/ GNSS). (ICRA2019, IEEE Spectrum (2019))

Tire Force Estimation of Wheeled Mobile Robots

We proposed a real-time tire forces and friction coefficient estimation technique using the constrained Kalman filter (TFF-CKF). The proposed TFF-CKF estimates the state and tire forces using onboard sensor measurements and selects the most proper friction coefficient using the Bayesian hypothesis selection algorithm. With the selected friction coefficient and tire model, the proposed TFF-CKF updates the tire forces using the constrained Kalman filter to enhance the robustness and accuracy. The proposed technique is verified by simulations and experiments, and the mean and standard deviation of the estimated friction coefficient are, respectively, 0.8307 (typical asphalt: 0.8-1.0) and 0.0198. (IROS2018)

Camera-GNSS-IMU Sensor Fusion

GNSS-IMU based sensor fusion is widely used for autonomous flying, which yet suffers from the inaccuracy and drift of the GNSS signal and also the failure with the loss of GNSS (e.g., indoor flying). For this, we propose a new framework for camera-GNSS-IMU sensor fusion, which, by fusing monocular camera information with GNSS and IMU, can improve accuracy and robustness of the autonomous flying. For camera and GNSS-IMU calibration, a new Kalman filter is also proposed, which runs in parallel with state estimation EKF and also utilize multiple keyframes generated from the camera information. (DroneCAN2016)

SLAM-IMU Fusion for Indoor Flying

GNSS has been widely used for outdoor autonomous flying of drone, which yet becomes unavailable for indoor flying. For this, we implement EKF-based sensor fusion by fusing mono-camera, rangefinder and IMU (i.e., gyroscope, accelerometer and magnetometer). We particularly adopt ORB-SLAM for monocular camera information processing to achieve fast localization of the drone, which is crucial for autonomous flying control. The drone is controlled by our own backstepping flying controller. (RAS2014)

Formation Flying w/ GNSS-IMU Fusion

Outdoor formation flying of three custom-built drones in a circular shape. For this, we utilize EKF-based state estimation in SE(3), which uses the information of GNSS, IMU, and barometer, consisting of : (1) state propagation by using the gyroscope and accelerometer of the IMU; and (2) measurement update by using the compass, barometer and GNSS. The drones are also controlled by in-house built backstepping control law presented in (RAS2014).

Autonomous Dynamic Driving

Control framework to enable an nonholonomic WMR to autonomously drive fast enough, yet, still less than a certain threshold to prevent wheel slippage. For this, instead of Newtonian vehicle modeling, we adopt Lagrange-D’Alembert formulation to explicitly relate the system’s state/control with constraint force, so that we can predict/detect possibility of violating no-slip condition. We also experimentally show that our control law, albeit based on simple Coulomb friction model, can still allow for dynamic driving of a RC car with no slippage. (ICRA2014)

INRoL