Published papers

Model-based control is a crucial component of robotic navigation. However, it often struggles with entrapment in local minima due to its inherent nature as a finite, myopic optimization procedure. Previous studies have addressed this issue but sacrificed either solution quality due to their reactive nature or computational efficiency in generating explicit paths for proactive guidance. To this end, we propose a motion planning method that proactively avoids local minima without any guidance from global paths. The key idea is repulsive potential augmentation, integrating high-level directional information into the Model Predictive Path Integral control as a single repulsive term through an artificial potential field. We evaluate our method through theoretical analysis and simulations in environments with obstacles that induce local minima. Results show that our method guarantees the avoidance of local minima and outperforms existing methods in terms of global optimality without decreasing computational efficiency.

カメラ画像を用いたビジュアルオドメトリの精度向上のため，地形の種類ごとに異なる特徴点検出アルゴリズムを適用する手法を提案し，ロボットを用いたフィールド試験によって有用性を実証した．

不整地を走行する車輪型移動ロボットの車輪力学モデルとステアリング時の挙動解析モデルをそれぞれ提案した．

Extraterrestrial drilling is an important sampling method for understanding the formation of another celestial body and finding traces of life beyond earth. Currently, such tasks mainly rely on remotely controlled space robots. Because of limited energy available for space robots, it is highly demanded to improve the energy efficiency in the extraterrestrial drilling tasks. In this paper, we propose an algorithm called combined power map (CPM) method that can optimize extraterrestrial drilling strategies in order to improve energy efficiency. The algorithm comprehensively considers the effect of sampling positions, robot positions, and robotic arm configurations on drilling energy consumption and has the ability to manage the energy efficiency gain and corresponding risks. Simulations show that the CPM method can significantly reduce the drilling energy consumption and effectively limit accompanying risks.

This paper introduces a Hardware-In-the-Loop Simulation (HILS) framework for a wheeled mobile robot traveling on sandy terrain. The HILS is a hybrid method that incorporates an experimental approach into a numerical simulation. The HILS in our work employs a single-wheel testbed for the experimental setup where the traction force of the wheel on soft soil under various slip conditions is measured. The dynamics of the robot vehicle in response to the wheel traction force is calculated by a simple dynamics model based on Newton’s equation of motion. A key technique of the HILS developed in this work is the damping coefficient implemented in the dynamics model. To verify the fidelity and stability of the HILS, wheel driving tests with the HILS are conducted in the following process: first, a wheel driving data (wheel translational velocity, wheel traction force, and wheel slip) with various input parameters (damping coefficient and wheel angular velocity) is experimentally measured only using the wheel testbed. Subsequently, another wheel driving data is measured while the HILS calculates the dynamic response of the wheel in accordance with the corresponding external forces. The validity of the HILS is compared based on the difference between the actual and calculated wheel driving data. In this process, the value of the damping coefficient is determined such that the dynamic response well matches with the actual driving data. Furthermore, wheel driving data with external disturbances to the wheel is experimentally simulated using the HILS.

In this paper, steering characteristics of a rigid wheel (tire) on loose soil is investigated. Based on terra-mechanic analysis, the lateral force characteristics of a driving wheel is modeled as a function of slip ratio and slip angle. The model suggests that the lateral force decreases according to the increment of the slip ratio and increases according to the increment of the slip angle. Such characteristics are confirmed and evaluated by experiments using simulated lunar-surface soil, called Lunar Regolith Simulant. The proposed model is validated with the experimental results in reasonable precision. A model that properly predicts the lateral force will be useful for future practical issues, such as controlling the steering motion of a vehicle for following a desired trajectory in the operational phase, and also to compare the feasibility and/or stability of candidate steering maneuvers in the motion planning phase.

This paper describes a steering trajectory analysis for planetary exploration rovers. Analysis of the steering trajectory is an important issue in order to plan and control the motion maneuvers of a planetary rover. Since a rover might easily slip and skid on loose soil, it is difficult to estimate and analyze the motion of the rover or the behavior of its wheels. The authors developed a model that respects the dynamics of each wheel's slip and skid behavior in order to deal with the steering trajectory of the rover. The developed model is called All-Wheel Dynamics Model. In the all-wheel dynamics model, the behavior of each wheel on loose soil is modeled based on terramechanics. The steering trajectory of the rover is obtained by a numerical simulation using the wheel-and-vehicle dynamics model. The proposed model is able to correspond to the experimental results with reasonable precision.

The ability of mobile robots to generate feasible trajectories online is an important requirement for their autonomous operation in unstructured environments. Many path generation techniques focus on generation of time- or distance-optimal paths while obeying dynamic constraints, and often assume precise knowledge of robot and/or environmental (i.e. terrain) properties. In uneven terrain, it is essential that the robot mobility over the terrain be explicitly considered in the planning process. Further, since significant uncertainty is often associated with robot and/or terrain parameter knowledge, this should also be accounted for in a path generation algorithm. Here, extensions to the rapidly exploring random tree (RRT) algorithm are presented that explicitly consider robot mobility and robot parameter uncertainty based on the stochastic response surface method (SRSM). Simulation results suggest that the proposed approach can be used for generating safe paths on uncertain, uneven terrain.

In this paper, steering characteristics of an exploration rover on loose soil is studied. Analysis of the steering characteristics is a key to plan and control the motion trajectory of a rover. Traditionally, such analysis has been made based on a model called "bicycle model" In that model, a four-wheel car-like vehicle is approximated by a two-wheel bicycle-like vehicle with the fore-wheels and the rear-wheels paired. However, the bicycle model does not show a good performance when a vehicle travels off-road. In order to analyze the steering characteristics of a vehicle on loose soil, the authors develop a model that respects the dynamics of each wheel's slip and skid behavior. The developed model is called All-Wheel Dynamics Model. In the all-wheel dynamics model, the behavior of each wheel on loose soil is modeled based on terramechanics. The motion trajectory of the vehicle is obtained by numerical simulation using the wheel-and-vehicle dynamics model. The validity of the proposed model is examined by the experiments of a wheel and a vehicle using simulated lunar-surface soil. The experimental results show that the proposed model provides a better approximation than the traditional bicycle model.

This paper presents analysis of traction mechanics and control of a lunar/planetary rover based on the models obtained from terramechanics. A case study has been conducted for a rover test bed to negotiate a slope of loose soil such as regolith that covers most of lunar surface. The tire traction force is modeled as a function of the vertical load and slip ratio of the wheel. Bekker's terramechanic formulae are employed to derive an improved practical model that calculates net traction force, referred to as Drawbar Pull, with a reasonable precision. Experiments are carried out in two phases. First, the physical behavior of a wheel on loose soil is observed using a single-wheel test bed, then the empirical parameters of the tire and soil are identified. Second, the slope climbing capability is studied by using a rover test bed that has independently driven four wheels. The traction margin and slip margin are defined to be used in a traction control. In the slope experiment, it turned out that the climbing capability was saturated at 14 degrees due to the lack of enough driving torque in wheels. But theoretical investigation suggests that this is not the limitation of terrain trafficability and climbing capability can be improved by increased driving torque and proper load distribution.

This paper presents a tangle- and contact-free path planning (TCFPP) for a mobile robot attached to a base station with a finite-length cable. This type of robot, called a tethered mobile robot, can endure long-time exploration with a continuous power supply and stable communication via its cable. However, the robot faces potential hazards that endanger its operation such as cable snagging on and cable entanglement with obstacles and the robot. To address these challenges, our approach incorporates homotopy-aware path planning into deep reinforcement learning. The proposed reward design in the learning problem penalizes the cable-obstacle and cable-robot contacts and encourages the robot to follow the homotopy-aware path toward a goal. We consider two distinct scenarios for the initial cable configuration: 1) the robot pulls the cable sequentially from the base while heading for the goal, and 2) the robot moves to the goal starting from a state where the cable has already been partially deployed. The proposed method is compared with naive approaches in terms of contact avoidance and path similarity. Simulation results revealed that the robot can successfully find a contact-minimized path under the guidance of the reference path in both scenarios.

Planetary surface exploration rovers must accurately and efficiently predict their mobility on natural, rough terrain. Most approaches to mobility prediction assume precise a priori knowledge of terrain physical parameters, however in practical scenarios knowledge of terrain parameters contains significant uncertainty. In this paper, a statistical method for mobility prediction that incorporates terrain uncertainty is presented. The proposed method consists of two techniques: a wheeled vehicle model for calculating vehicle dynamic motion and wheel-terrain interaction forces, and a stochastic response surface method (SRSM) for modeling of uncertainty. The proposed method generates a predicted motion path of the rover with confidence ellipses indicating the probable rover position due to uncertainty in terrain physical parameters. Rover orientations and wheel slippage are also predicted. The computational efficiency of SRSM as compared to conventional Monte Carlo methods is shown via numerical simulations. Experimental results of rover travel over sloped terrain in two different uncertain terrains are presented that confirms the utility of the proposed mobility prediction method. ©2010 IEEE.