This paper focuses on the integration of inertial measurement unit (IMU) with two real-time kinematic global positioning system (GPS) units in an adaptive Kalman filter (KF) for driftless estimation of a vehicle's attitude and position in 3-D. The observability analysis reveals that 1) integration of a single GPS with IMU does not constitute an observable system; and 2) integration of two GPS units with IMU results in a locally observable system provided that the line connecting two GPS antennas is not parallel with the vector of the measured acceleration, i.e., the sum of inertial and gravitational accelerations. The latter case makes it possible to compensate the error in the estimated orientation due to gyro drift and its bias without needing additional instrument for absolute orientation measurements, e.g., magnetic compass. Moreover, in order to cope with the fact that GPS systems sometimes lose their signal and receive inaccurate position data, the self-tuning filter estimates the covariance matrix associated with the GPS measurement noise. This allows the KF to incorporate GPS measurements in the data fusion process heavily only when the information received by GPS becomes reliably available. Finally, test results obtained from a mobile robot moving across uneven terrain demonstrate driftless 3-D pose estimation.

Exploration of unknown planets using autonomous rovers requires an efficient onboard localization system capable of estimating precisely rover position and orienta tion. This paper presents a method to refine rover odometry using an Iterative Closest Point (ICP) algorithm applied on 3-D LIDAR panoramas collected by a rover. Our approach takes two LIDAR scans from different locations, subsamples and simplifies them. Then it performs a registration which gives a 6 degrees of freedom rigid body transformation representing the estimation of rover odometry error accumulated during the rover motion from the first scanned location to the other. The proposed LIDAR ICP Pose Refiner (LIPR) can perform registrations in the presence of large misalignments and with overlap ratios below 50%. An intensive benchmark test using field data has shown that LIPR can tolerate larger error than three standard approaches. The paper also reports experimental results of rover pose refinement tests performed at the Mars Emulation Terrain (MET) of the Canadian Space Agency (CSA). LIPR has been tested online at more than 174 different locations of MET, keeping odometry error on the order of 1% of distance traveled.

The embodiment design of a two-wheeled self-balancing human augmentation robot for the mobility-challenged is reported. The prototype relies on a dead-reckoning multisensory system consisting of i) two optical incremental encoders and ii) a solid-state tilt sensor. The command inputs are provided to the robot controller, which is based on PC/104 technology, by means of a RF control unit. After describing the research motivation and application of the system, a set of robot design solutions is outlined along with technical discussions on component layout, payload holder and chassis design issues. A few simulation results on the motion control performance are included as well

A real-time kinematic (RTK) global positioning system (GPS) has been identified for potentially being used as a ground-truth sensor for testing robotic rovers for planetary exploration. A series of environmental tests needs to be performed in order to validate the performance of the sensor at hand before being used as a ground-truth system. This paper focuses on the performance evaluation of the RTK GPS at Axel Heiberg Island Canadian Space Agency’s Analogue Research Network (CARN) site. This is one of the officially recognized terrestrial analogues, that is places on Earth that approximate the geological, environmental and putative biological conditions on Mars and other planetary bodies (Hipkin et al.). The challenge lies in the use of the equipment at Arctic latitudes. The results show that the system performed according to specifications even in this challenging environment.

This paper focuses on the design and test results of an adaptive variation of Kalman filter (KF) estimator based on fusing data from Inertial Measurement Unit (IMU) and two Real Time Kinematic (RTK) Global Positioning Systems (GPS) for driftless 3-D attitude determination and robust position estimation of mobile robots. GPS devices are notorious for their measurement errors vary from one point to the next. Therefore in order to improve the quality of the attitude estimates, the covariance matrix of measurement noise is estimated in real time upon information obtained from the differential GPS measurements, so that the KF filter continually is "tuned" as well as possible. No a priori knowledge on the direction cosines of the gravity vector in the inertial frame is required as these parameters can be also identified by the KF, relieving any need for calibration. Next, taking advantage of the redundant GPS measurements, a weight least-squares estimator is derived to weight the GPS measurement with the "good" data more heavily than the one with "poor" data in the estimation process leading to a robust position estimation. Test results are presented showing the performance of the integrated IMU and two GPS to estimate the attitude and location of a mobile robot moving across uneven terrain.

Reported in this paper are the implementation and testing of the real-time control of a two-wheeled mobile robot. The robot is underactuated, its mobility and control inputs being three and two, respectively. The control challenge faced here is to reduce the oscillations of the intermediate body while following a desired path. A controller, introduced elsewhere, is implemented using a real-time operating system on a novel control architecture. The control algorithm was tested using three different test motions: translational, rotational, and uphill.

This paper focuses on the experiments conducted using a telerobot for the augmentation of wheelchair users. After providing the motivation and the background material, a strawman task is formulated. A robot is then conceived to meet the assigned task (i.e. user, environment and payload definitions). The proposed robot meets both cost and control simplification requirements necessary to the success of a robotic assistive device. A minimalistic design allows to achieve the requirements on cost and control complexity. An architecture based on a minimum number of driving units and sensors is devised. Experiments on the interactive control of the robot are performed. We demonstrate that the robot is capable to navigate through a cluttered environment while being teleoperated. Experiments also show that the system remains in its footpring when a rotation in place is assigned by the user; this is an important feature that prevents the system from colliding with any object nearby. Finally, always via experiments, we show that the system is capable to bring a tray of drinks, food and reading material while being teleoperated.

This work deals with the robustness and controllability analysis for autonomous navigation of two-wheeled mobile robots. The analysis of controllability of the systems at hand is conducted using both the Kalman rank condition for controllability and the Lie Algebra rank condition. We show that the robots targeted in this work can be controlled using a model for autonomous navigation by means of their dynamics model: kinematics will not be sufficient to completely control these underactuated systems. After having proven that these autonomous robots are small-time locally controllable from every equilibrium point and locally accessible from the remaining points, the uncertainty is modeled resorting to a multiplicative approach. The dynamics response of these robots is analyzed in the frequency domain. Upper bounds for the complex uncertainty are established.

This paper introduces a new family of two-wheeled mobile robots. The mathematical model of the proposed robots is formulated along with a controllability analysis, also taking into account the oscillations of the intermediate body. We prove that it is possible to completely control these robots using only the wheel motors, while tracking a desired path

Aspects of the mechatronic design and prototyping of a robot designed and realized with the quasiholonomy property are discussed here. After an overview of the motivation and applications of the project at hand, we show how we faced the challenges encountered in the mechatronic design and implementation of the prototype. The actuation system and power supply dimensioning are then described, along with the selection of the on-board control unit and programming of the real-time operating system. Advantages of quasiholonomic robotic systems are substantiated

We report the analysis and simulation of the dynamical behaviour of a novel two-wheeled mobile robot, its novelty lying in the ease with which it can be controlled, as a result of its quasiholonomy, a concept introduced elsewhere. The mathematical model of the robot is formulated in the framework of the Lagrange formalism, by means of an orthogonal complement and the holonomy matrix, concepts introduced in previous works. A few simulations for the validation of the aforementioned model have been conducted. Moreover, we provide the dynamic response to different inputs and different initial conditions, which is crucial for the design and control of the robot.

We report on the control of semi-autonomous two-wheeled mobile robots undergoing large variations of the payload. These robots being underactuated, their control poses some challenges when resorting to a simple controller. The latter being a linear controller, its robustness with respect to model uncertainty needs to be investigated. To do this, we conducted a time-domain analysis of robustness of the foregoing controller with respect to parametric and unmodeled dynamics uncertainty. The controller, which is designed by a dominant second-order pole technique, turns out to be fragile (as opposed to robust) with respect to unmodeled dynamics uncertainty. In order to cope with this problem, a linear-quadratic regulator is designed. By numerical simulation we show that the latter features robustness with respect to both types of uncertainty.

The robust design of a novel mobile robot, which comprises two driving wheels and an intermediate body carrying the payload, is the subject of this paper. We prove that, by virtue of the robot architecture, the kinetostatic model of the system is isotropic. Moreover, regarding the robot dynamic response, a robust design problem is formulated by minimizing the design bandwidth of the generalized inertia matrix of the robot over its architecture parameters. Furthermore, design conditions are given for the robot performance in trajectory-tracking to be feasible. Finally, a numerical comparison of two design solutions, one feasible and one robust, is provided by means of simulation runs. We demonstrate that the robust design solution doubles robot performance in trajectory-tracking, while reducing the oscillations of the intermediate body, by 40%, when compared with the feasible solution.

We report on the controllability of a novel mobile robot which comprises two driving wheels and an intermediate body carrying the payload. By virtue of quasiholonomy, a concept introduced elsewhere, the robot is underactuated by design. One challenge here is the control of the motion of the intermediate body, which will tend to rotate about the wheel axis as the wheels are actuated. We prove that it is possible to completely control the robot using only the wheel motors, while tracking a desired trajectory, with apparent advantages in terms of cost, weight and efficiency. To do this, we show that every linearization of the robot dynamics model around an equilibrium point verifies the Kalman rank condition for controllability. Moreover, using modern results from nonlinear control theory, we prove that the robot is locally accessible and small-time locally controllable.

We report the holonomy analysis of three wheeled mobile robots. Two of these, Vuton II and Nomad, have full mobility and are claimed by their inventors to be holonomic. We show that, in fact, they are nonholonomic. Nevertheless, conditions are given under which these robots can be designed for quasiholonomy, a concept introduced elsewhere and recalled here. The third robot, Quasimoro, has a mobility of two and is unconditionally quasiholonomic. It is shown that if the wheel mechanism of a robot with full mobility is designed in such a way that the platform is much heavier than the wheel mechanism, then the underlying mathematical models are free of the nonholonomy term.