In this study, we propose a method to simultaneously reduce disparity errors at long distances and practical operation-induced errors in distance measurement using a low-cost stereo camera and YOLO-based object detection. By leveraging the characteristics of an omnidirectional mobile robot, the target object is positioned at the center of rotation, and the robot actively performs rotational motion. The horizontal pixel shift ΔX between frames before and after rotation is observed. Based on the pinhole camera geometry model, we derive a correction formula to inversely calculate the true distance D from the initial measured distance Destand the rotation angle θ. Experimental results show that the initial measurement error was reduced to 4% and 11% at true distances of 1.0 m and 1.5 m, respectively, with optimal correction observed for θ = 10° to 30°. Furthermore, we propose that a hybrid approach, using the initial measure as-is in close-range scenarios, is effective. The proposed method, which is applicable in real time, contributes to improving depth measurement accuracy for autonomous mobile robots in low-cost environments.

This paper presents a large-scale omnidirectional vehicle with six active casters. The vehicle can be divided into three segments—front, mid, and rear—with the front and rear segments capable of functioning as independent omnidirectional vehicles, each with its own control and drive system. These three segments can be coupled to carry large and heavy loads of 1000-2000 kg. The vehicle also features a reconfiguration function that adjusts the wheel layout, for example, to maintain stability. We describe control methods for both cooperative and independent operation in coupled and decoupled states, including layout adjustment.

In countries with aging societies and decreasing working population, the demand for collaborative robots is increasing. To realize safe and flexible human–robot collaboration, force control is essential. However, torque sensors are limited by the frequency bandwidth and installation space. This study developed a torque estimation method that does not use torque sensors and instead focuses on the nonlinear torsional stiffness characteristics of a reduction gear. By utilizing the frequency characteristics of an actuator, which change according to the applied torque, a nonlinear estimation model was constructed using the gain and phase characteristics at specific frequencies. In addition, a frequency selection procedure for estimation was developed based on indices derived from the continuity and sensitivity of the frequency characteristics. Experimental results confirmed that high estimation accuracy can be achieved in medium- and high-torque regions. Future work will focus on improving estimation accuracy in low-torque region and extending the method to achieve real-time estimation.

As technology continues to advance, the realization of autonomous driving is gradually transitioning from a dream to a tangible reality. However, achieving the functionality of autonomous vehicles requires accurate self-localization capabilities and precise perception and understanding of the surrounding environment. Among various positioning technologies used in autonomous driving systems, Global Navigation Satellite Systems (GNSS) and Light Detection and Ranging (LiDAR) are commonly employed techniques, each with its own advantages and limitations. Integrating these two technologies and exploring their flexible transition is a necessary research area. This study focuses on integrating the strengths of GNSS and LiDAR to achieve smooth transition in different scenarios, thereby improving the real-time accuracy of localization for autonomous vehicles in complex environments. To accomplish this, we propose a “Smooth Move Algorithm” that ensures seamless transitions between GNSS and LiDAR, enhancing the diversity, flexibility, and precision of the positioning system. This research aims to enhance and improve autonomous driving localization technology, contributing to the establishment of safer and more efficient intelligent transportation systems. Moreover, it provides a promising approach for integrating GNSS and LiDAR in novel ways, opening up new possibilities for the development of intelligent transportation systems and facilitating the widespread application of autonomous vehicles.

As the impact of traffic accidents caused by older adults is significantly large, they are often advised to surrender their driving license. On the contrary, the decrease in the number of human drivers for public transportation is a serious problem in rural areas where populations are aging and declining. With a rapidly aging society, mobility-related problems among the elderly have gained significant attention. To address these issues, dedicated autonomous public transportation is essential, thereby necessitating field-operational tests involving autonomous buses. However, the deployment of autonomous bus-related field-operational tests in Japan is relatively slow as compared to the that in the U.S. and China. This is attributed to the few by-wire controllable bus systems that meet the requirements of Japanese vehicular laws for field-operational tests. To resolve these issues, we propose an autonomous bus using existing joystick driving system solutions. The steering wheel and gas/brake pedals are activated using electric motors of the joystick driving system, thus achieving drive-by-wire. Therefore, we selected this joystick-driving system attached to a bus as the basis for an autonomous bus, to meet the requirements of Japanese vehicular laws. Systems including LIDAR, GNSS, gyro sensors, and cameras enable the bus to be automatically controlled using the proposed driving system. This proposed prototype bus has passed vehicular inspection, and field-operational tests have been performed on public roads using this bus. The results of these tests confirm the viability of our proposed solution.

At present, the aging rate in Japan is among the highest in the world, and it is progressing at such a rate that it cannot be seen in other countries. Therefore we consider that the number of persons who are difficult to move by oneself will increase. We focus on an electric wheelchair. There are classified roughly into two types by the difference of installed position of drive wheels, and it results in the difference in run properties. We pay our attention to low straightness of front-wheel-drive under these properties. This research aims at finding the cause of low straightness and proposition of the method which improves straightness. We built a human - wheelchair model and performed simulations, and we performed the automatic drive experiments with the wheelchair equipped with a laser sensor. Based on these results, we propose a control method to improve straightness of a front-wheel-drive wheelchair. Finally we confirmed the usefulness of our method by simulations, and experiments.

In this paper, a new type of an active-caster with a single ball transmission (ACROBAT-S) is proposed. The proposed ball transmission is a novel mechanism which realizes combining of two motor powers together to rotate the single ball in a 2Dway and simultaneously distributing the combined power to a wheel shaft and a steering shaft of an active-caster in an appropriate ratio. The transmission design enables to remove a sensor for detecting the wheel orientation for coordinated control. Also, it would be possible to build an omnidirectional robot with three active casters to be controlled its 3D motion by three motors with no redundancy. The new concept of ACROBAT-S gives many advantages compared with the original ACROBAT since the number of balls is reduced from two to one, therefore the number of friction drive between a ball and rollers or a ball to a ball is also reduced from five to three. These features contribute not only to simplify the mechanism design, but also to enhance its performance since slippages and energy losses would be reduced. To verify the design concept, we derive and analyze a kinematic model of a single ball transmission mechanism. Furthermore the motions of ACROBAT-S are verified by Solid Works 3D simulator. From the results, it is confirmed that the proposed single ball transmission is applicable for an omnidirectional wheel mechanism to realize omnidirectional motions.

A new type of the active-caster robotic drive with dual-wheels and differential mechanism (ACRO-DD) is presented in this paper. The active-casters have been used for the driving devices for wheeled mobile robots, wheelchairs, etc. The new active-caster equips dual-wheels which are driven by a wheel motor via a differential gear mechanism. The dual-wheel configuration contributes to reduce frictions in turning, which enables smooth caster motions. For maintaining the two-wheel-contact to the ground even on the irregular surface, an individual suspension is installed on each wheel. The dual-wheel-design enables the mechanism to equip large wheels compared with the single wheel design, which realizes high step climbing capability. Moreover, to realize a mechanism with compact size and light weight, we install a novel sensor system for detecting the absolute orientation of the wheel using two smallpotentiometers. The mechanical design and the sensing system are embodied by a prototype which shows expected caster motions with high mobility.

We have previously developed a five-wheeled wheelchair with an active-caster drive system. To improve motion performance of the wheelchair, we have proposed the function which two types of motions, such as rear-like drive and front-like drive mode, can be selected by using the linear actuator. In this paper, we propose a step climbing strategy and develop a mechanism. The proposed five-wheeled wheelchair has five points of contact to the ground in the stand condition. To make the front casters of the wheelchair to be hovered from the ground, a “static wheelie” configuration in which large wheels and a drive wheel touch to the ground is taken by maintaining the static stability of a wheelchair. We propose a novel and unique strategy by utilizing the statically stable wheelie configuration. Additionally, the drive wheel is lifted onto the step by cooperative control between the linear actuator and the drive wheel. We confirm that the proposed method is applicable by comparing it with a step climbing method that uses only the traction force of the drive wheel. To achieve the proposed step climbing method, we develop a novel link mechanism that can perform not only the step climbing method but also the functions mentioned above and confirm the design conditions. After the prototype is built, we confirm that a user can climb over a step by the proposed method and mechanism.

This paper presents a steering-wheel control system with a joystick interface for stable car driving at a high speed. The joystick drive system allows a handicapped person to drive a car using a joystick interface with a small force and short stroke. However when a vehicle running at a very high speed, a wrong joystick operation, such as a sudden and wide steering, results in a car accident quite easily. The additional problem is that a car could not be controlled stable if joystick sensitivity is not appropriate especially at high speed. The other hand when parking a car, a large steering angle provides easy parking maneuvers to a driver. To realize these multiple requirements and solve some problems for the steering control, we propose a steering control system with a steering angle restriction and a variable sensitivity function based on vehicle velocity information. In this paper, some types of variable sensitivity methods are proposed and one of them is selected through driving tests by a human driver. By the variable sensitivity function, approx. 85km/h driving is realized.

In this paper, we study the design of omnidirectional mobile robots with Active-Caster RObotic drive with BAll Transmission (ACROBAT). ACROBAT system has been developed by the authors group which realizes mechanical coordination of wheel and steering motions for creating caster behaviors without computer calculations. A motion in the specific direction relative to a robot body is fully depends on the motion of a specific motor. This feature gives a robot designer to build an omnidirectional mobile robot propelled by active-casters with no redundant actuation with a simple control. A controller of the robot becomes as simple as that for omni-wheeled robotic bases. Namely 3DOF of the omnidirectional robot is controlled by three motors using a simple and constant kinematics. ACROBAT includes a unique dual-ball transmission to transmit traction power to rotate and orient a drive wheel with distributing velocity components to wheel and steering axes in an appropriate ratio. Therefore a sensor for measuring a wheel orientation and calculations for velocity distributions are totally removed from a conventional control system. To build an omnidirectional vehicle by ACROBAT, the significant feature is some multiple drive shafts can be driven by a common motor which realizes non-redundant actuation of the robotic platform. A kinematic model of the proposed robot with ACROBAT is analyzed and a mechanical condition for realizing a non-redundant actuation is derived. Based on the kinematic model and the mechanical condition, computer simulations of the mechanism are performed. A prototype two-wheeled robot with two ACROBATs is designed and built to verify the availability of the proposed system. In the experiments, the prototype robot shows successful omnidirectional motions with a simple and constant kinematics based control.

This paper presents a new active-caster drive system for omnidirectional mobile robots. The active-caster presents omnidirectional motion on the ground with providing an active traction force in an arbitrary direction by controlling a 2DOF planar velocity vector created by independent two motors. The proposed mechanism includes a dual-ball transmission to transmit traction power to rotate and orient a drive wheel with distributing velocity components to wheel and steering axes in an appropriate ratio. One of the significant features of the proposed mechanism is that a velocity ratio between wheel rotation and steering rotation is determined mechanically by the dual-ball transmission, where the ratio is a function of an orientation of a wheel, therefore a sensor for measuring a wheel orientation and calculations for velocity decomposition are totally removed from a conventional control system.

This paper presents mechanism and control of a four-wheel-drive (4WD) robotic platform for wheelchairs. The 4WD mechanism equips four wheels, two omni-wheels in front and two normal tires in rear. The normal wheel and the omni-wheel, mounted on the same side of the base, are interconnected by a synchro-drive transmission to rotate in unison with a drive motor. To control chair orientation on the 4WD platform, the third motor is installed on the platform. The chair with the proposed omnidirectional 4WD system is capable of moving in any direction which is so-called holonomic and omnidirectional mobility. The holonomic omnidirectional mobile capability enables a person to drive a mobile system with no knowledge about the drive mechanism or its configuration since it can move in any direction and rotate from any configuration of the mechanism. This paper presents the mechanism and control of 4WD platform to which powered-caster omnidirectional control is applied. The rotations of two pairs of wheel and vertical axis of a chair are controlled in such a way that it performs as a powered twin-caster to realize holonomic motions. The prototype wheelchair is designed and built to verify the mechanism design and control method. The smooth and flexible onidirectional motions are presented by the experiments.

This paper presents a joystick car drive system for a handicapped person using a wheelchair. The joystick drive system enables handicapped person to drive a car with his/her single hand. The joystick operation in back and force direction controls acceleration or deceleration of a car while that in left and right direction turns a steering wheel. Therefore a person, who has disabilities in legs together with some disabilities such as not enough force nor move their arms in wide area, can drive a car by oneself. Additionally, a wheelchair driver does not have to change seat from a wheelchair to a car seat. The developed van equips with a lift on the back and wheelchair user can access to the drivers position with propelling a wheelchair inside of a van. For maintaining driving safety of a van in case of system fault, battery fault, etc., gas and brake pedals are physically moved by mechanical linkage which is connected to a joystick lever. Therefore a van can be stopped by manual operation in any case. A steering wheel is driven by an electric motor which is controlled by a micro-computer system. A steering angle is controlled by PID feedback to track the reference angle given by a joystick. To realize the joystick drive system, we design a steering drive mechanism which includes a DC motor, a magnetic clutch, and a potentiometer for steering angle detection, and transmission gears. Most of cars manufactured in these days equip with power steering systems which reduce steering operation power of a human driver. Therefore the developed system can utilize the power steering system for realizing the joystick car drive system with minimum capacity of a motor. The prototype mechanism is mounted on a actual van type car and tested a realizability of the proposed joystick car drive system.

This paper presents analysis and experiments of a 4WD omnidirectional wheelchair for climbing a high step. For enhancing the mobility and maneuverability of standard wheelchairs, a new type of omnidirectional mobile platform with four-wheel drive (4WD) mechanism is introduced. The mobile platform equips four wheels, two omni-wheels in front and two normal tires in rear. The normal wheel and the omni-wheel, mounted on the same side of the base, are interconnected by belt transmissions to rotate in unison with a drive motor, i.e. a synchro-drive transmission. To rotate a chair at the center of the mobile base about vertical axis, the third motor is installed on the platform. The prototype wheelchair with the proposed 4WD system is capable of moving in any direction and climbing over a high single step. The conventional 4WD system provides enhanced step climb capability. However the maximum step height is restricted by the rear wheels. Namely, front wheels can climb up higher step than the rear wheels although the both wheels are identical in diameter. To clarify the difference between front and rear wheels in the step climb capability, the statics of the 4WD system is analyzed. It is suggested that the reaction of a drive torque cause the front-rear differences and its influence depends on a wheelbase and a vehicle weight distribution ratio between front and rear wheel axes. For varying a vehicle weight distribution ratio, we develop a chair tilting mechanism for the 4WD wheelchair with a linear drive mechanism controlled by a stability algorithm using an inclinometer. To verify the analysis results and availability of the chair tilting mechanism, step climb capabilities of a 4WD wheelchair prototype is tested in experiments. In the experiments, both front and rear wheels of the prototype wheelchair with a human can successfully climb a 90 mm step while the rear wheels fail when the chair tilting system is disabled.

This paper presents a new type of omnidirectional and holonomic mobile platform with a four-wheel-drive (4WD) mechanism for improving traction of electric wheelchairs on slippery surfaces and enhancing mobility on rough terrain. The 4WD mechanism includes a pair of normal wheels on the rear and a pair of omniwheels on the front. The normal wheel in rear and the omniwheel in front on the same side of the drive mechanism are connected by a power transmission to rotate in unison with a common motor. Omniwheels enable the front of the mechanism to roll freely from side to side. A third motor turns the chair about a vertical axis at the center of the mobile platform Thus the platform is controlled by three motors, no redundancy and no motor confliction are found on the drive system High step climb capability of the 4WD mechanism solves the insufficient mobility problem and the omnidirectional control enables simple and easy operation with no complicated maneuvering. To verify the availability of the proposed mobile system for wheelchairs, a prototype is designed and built. The fundamental step climb capability and holonomic and omnidirectional mobility are verified by a series of experiments.

This paper presents a new modeling method for holonomic omnidirectional mobile robots propelled by powered-casters. A mobile robot with powered-casters is redundantly actuated since it is controlled by four or more numbers of motors while its configuration is represented by 3DOF on the ground. The redundant architecture leads inverse problems on motion planning, actuator coordination, static or dynamic control, and dead-reckoning calculations with redundant wheel information. To overcome those problems led by the redundant architecture, a new modeling method based on the virtual linkage concept is proposed. The virtual linkages are applied for representing virtual deformations of a robot frame even though an actual robot frame cannot be deformed. The virtual frame deformation provides additional DOF on the robot, which allows total robot DOF equal to the number of actuators equipped with the mobile robot. The acquired n-wheeled robot model with virtual linkages satisfies a class of 2n by 2n system, no redundancy is found in a control system where no inverse calculations of non-square matrices, nor motion planning of redundant actuators. The proposed virtual link model is generalized and a unified procedure is presented for n-wheeled mobile robots. A modeling example is shown and experiments of a three-wheeled robot prototype are performed for evaluating the proposed model and its implementation to a redundantly actuated omnidirectional mobile robot

A new type of mobile platform with a novel wheel mechanism for holonomic and omnidirectional robots is presented. The drive mechanism equips with a couple of standard wheels in a differential-drive manner driven by two independent motors. A steering joint, driven by the third motor, is mounted on the differential drive mechanism at a location off-centered from a mid-point of the two wheels. Thus, the drive mechanism has a dual-wheel-caster configuration. First, the concept of the original caster drive is reviewed briefly followed by kinematic analysis of the dual-wheel-caster mechanism. A prototype vehicle is designed, built and its mobility tested. Other possible variation for the future vehicle design with a traditional synchro-drive mechanism is proposed. The kinematics and statics of the drive mechanism are also analyzed. All wheel driving without over constraint decoupled control and efficient driving can be achieved with the proposed synchro-drive.

A reconfigurable mechanism for varying the footprint of a four-wheeled omnidirectional vehicle is developed and applied to wheelchairs. The variable footprint mechanism consists of a pair of beams intersecting at a pivotal point in the middle. Two pairs of bad wheels at the diagonal positions of the vehicle chassis are mounted, respectively, on the two beams intersecting in the middle. The angle between the two beams varies actively so that the ratio of the wheel base to the tread may change. Four independent servo motors driving the four ball wheels allow the vehicle to move in an arbitrary direction from an arbitrary configuration as well as to change the angle between the two beams and thereby change the footprint. The objective of controlling the beam angle is threefold. One is to augment static stability by varying the footprint so that the mass centroid of the vehicle may be kept within the footprint at all times. The second is to reduce the width of the vehicle when going through a narrow doorway. The third is to apparently change the gear ratio relating the vehicle speed to individual actuator speeds. First the concept of the varying footprint mechanism is described, and its kinematic behavior is analyzed, followed by the three control algorithms for varying the footprint. A prototype vehicle for an application as a wheelchair platform is designed, built, and tested.

A new reconfigurable mechanism for varying the footprint of a four-wheeled omnidirectional vehicle is presented. A chair is mounted on a moving platform with four ball-wheels allowing holonomic, omnidirectional motion. Two pairs of ball-wheels at the diagonal positions of the chassis are mounted, respectively, on two beams that intersect at a pivotal point in the middle. The angle between the two beams crossing at the pivotal point is varied actively so that the ratio of the wheel base to the tread may change. Four independent servomotors driving the four ball-wheels allow the vehicle to exhibit holonomic motion. The beam angle is controlled based on the measurement of the mass centroid position in such a way that the mass centroid can be kept within the footprint at all times. The concept of the reconfigurable footprint mechanism is described, and its kinematics is analyzed followed by the control design for augmenting vehicle stability. A prototype vehicle for wheelchair platform application is designed.

A new type of holonomic omnidirectional vehicle which consists of wheel mechanisms without free rollers but with conventional tires is presented. We first present the concept of a new driving wheel mechanism with a tire in the conventional style such as a rubber tire or a pneumatic tire, an "offset steered driving wheel". The configuration of the wheel has an offset distance between the wheel axle and the steering axle along the direction of wheel travelling. We then describe the kinematics and the control method of the offset steered driving wheel to provide an omnidirectional and holonomic capability for a vehicle by actuating the wheel axis and the steering axis independently. The wheel is a nonholonomic system, and we introduce a control strategy to avoid a nonholonomic constraint. Simulation results of the single driving wheel and the two-wheeled vehicle are performed to test the control method. The concept of offset steered driving wheel is implemented and tested on a prototype of the holonomic omnidirectional vehicle. Experimental results demonstrate high mobility of the vehicle prototype.

An omnidirectional holonomic mobile robot assembled with nonholonomic wheels has been proposed. The driving wheel axis of the nonholonomic wheel has an offset distance from the steering axis of the wheel. A wheel of this style is representative of the nonholonomic system and is conventionally used for a caster wheel. In the case of the proposed mobile robot, a wheel of this style is used as a driven wheel by actuating a wheel axle and a steering axle independently. First, we describe the modeling and control of the single nonholonomic wheel, and then describe the holonomic vehicle which has a couple of these nonholonomic wheels. To avoid the nonholonomic characteristics of the wheel, we have introduced a control method which restricts the controlled statement to only the translational velocity of the supporting point of the wheel. Computer simulations of the single wheel and of the two wheeled vehicle, and experimental results of the single wheel have shown good performances.