University of Technology, Sydney. Faculty of Engineering. / The thesis focuses on issues of vehicle modelling incorporating wheel-terrain interaction and low-level control design taking into account uncertainties and input time delay. Addressing these issues is of significant importance in achieving persistent autonomy for outdoor UGVs, especially when navigating on unprepared terrains. The test-bed vehicle used for this research is retrofitted from an all-terrain 20-hp, 0.5-tonne vehicle. Its driveline system consists of an internal combustion engine, continuous variable transmission (CVT), gearbox, differential, chains, and eight wheels. The vehicle is driven in the skid-steering mode, which is popular for many off-road land-vehicle platforms. In this thesis, a comprehensive approach is proposed for modelling the driveline. The approach considers the difference in speed between two outputs of the differential and the turning mechanism of the vehicle. It describes dynamics of all components in the vehicle driveline in an integrated manner with the vehicle motion. Given a pattern of the throttle position, left and right braking efforts as the inputs, the dynamic behaviour of the wheels and other components of the UGV can be predicted. For controlling the vehicle at the low level, PID controllers are firstly used for all actuators. As many components of the vehicle exhibit nonlinearities and time delay, the large overshoots encountered in the outputs can lead to undesirable vehicle behaviours. To alleviate the problem, a novel control approach is proposed for suppression of overshoots resulting from PID control. Sliding mode control (SMC) is employed, for this, with time delay compensated by using an output predictor. As a result, the proposed approach can improve significantly system robustness and reduce substantially step response overshoot. Notably, the design is generic in that it can be applied for many dynamic processes. Knowledge of the interaction between the UGV and the terrain plays an important role in increasing its autonomy and securing the safety for off-road locomotion. In this regard, vehicle kinematic equations are combined with the theory of terramechanics for dynamic modelling of the interaction between the vehicle wheels and a variety of terrain types. Also, a fast algorithm is developed to enable online implementation. The novel interaction model takes into account the relationship between normal stresses, shear stresses, and shear displacement of the terrain that is in contact with the wheels in deriving the three-dimensional reaction forces. Finally, all modelling and control algorithms are integrated into a unique simulator for emulating the vehicle mobility characteristics. In particular, the wheel’s slip and rolling resistance can also be derived to provide useful information for closed-loop control when the UGV is navigating in an unknown environment. The simulator, as a tool for analysing the vehicle mobility, is helpful for further research on relevant topics such as traction control, safe and effective locomotion.

The first part of this thesis documents experimental investigation into the use of vision for wheeled robot navigation problems. Specifically, using a video camera as a source of feedback to control a wheeled robot toward a static and a moving object in an environment in real-time. The wheeled robot control algorithms are dependent on information from a vision system and an estimator. The vision system design consists of a pan video camera and a visual gaze algorithm which attempts to search and continuously maintain an object of interest within limited camera field of view. Several vision-based algorithms are presented to recognize simple objects of interest in an environment and to calculate relevant parameters required by the control algorithms. An estimator is designed for state estimation of the motion of an object using visual measurements. The estimator uses noisy measurements of relative bearing to an object and object's size on an image plane formed by perspective projection. These measurements can be obtained from the vision system. A set of algorithms have been designed and experimentally investigated using a pan video camera and two wheeled robots in real-time in a laboratory setting. Experimental results and discussion are presented on the performance of the vision-based control algorithms where a wheeled robot successfully approached an object in various motions. The second part of this thesis investigates the coordination problem of flocking in multi-robot system using concepts from graph theory. New control laws are presented for flocking motion of groups of mobile robots based on several leaders. Simulation results are provided to illustrate the control laws and its applications.

Insects are confronted with the problem of locating food, mates, prey and hosts for their young over long distances, which they often overcome using chemical plume tracking. Tracking a plume of chemical back to its source is made difficult due to the complexity of plume structure. Turbulence and shifts in the wind direction prevail over diffusion in the spreading of an airborne chemical from a point in most cases, producing intricate plumes consisting of filaments of high chemical concentration interspersed with regions of clean air. It has been proposed that insects achieve plume tracking in this environment through variations of anemotaxis, which involves travelling upwind when an attractive chemical is perceived. This study aimed to investigate anemotaxis through the use of a mobile robot to test the efficacy of algorithms which mimic the way insects achieve plume tracking and also to determine whether these algorithms are an effective means of plume tracking for a mobile robot under a range of conditions. To achieve the aims of this study, various plume-tracking algorithms were implemented on a mobile robot built to model a plume-tracking insect and their performance was compared under a range of wind conditions. The algorithms tested were based upon a range of plume-tracking hypotheses. The simplest algorithm was surge anemotaxis, where the robot surged upwind in the presence of an attractive chemical and performed crosswind casting (back and forth motion) in the absence of chemical. The other algorithms tested were the counterturner, where the robot zigzagged upwind, and two bounded search methods. To allow these algorithms to be appropriately implemented, a robot model was constructed that could move in two dimensions and sense the wind velocity and ion level at a point in space. An ion plume was used instead of a chemical plume in each test as it behaves in a similar manner to a chemical plume, but ion sensors have response and recovery times far more rapid than conventional chemical sensors, similar to insects. The plume-tracking robot was tested in three series of tests. Initially, the entire range of plume-tracking algorithms was tested in a wind tunnel with fixed wind direction for a range of wind speeds and release positions. The second series of tests compared the performance of the surge anemotaxis and bounded search algorithms, again in a wind tunnel, but with a wind shift of 20° during some of the tests. The algorithms were tested with and without a direct crosswind surge response to detected wind shifts. The third set of tests examined the performance of the simple and wind shift response algorithms outdoors using natural wind to produce the plume. All algorithms tested achieved successful plume tracking in some conditions. The surge anemotaxis and triangular bounded search algorithms were particularly successful. The tests also showed that the paths obtained from tests undertaken in natural outdoor wind conditions varied greatly from those undertaken in a wind tunnel. This indicates the need to test plume-tracking algorithms in natural environments. This is vital both in the investigation of insect plume-tracking behaviour, as insects navigate in these environments, and in the process of producing plume-tracking robots that are capable of operating effectively in these conditions. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=1287973 / Thesis (Ph.D.) -- University of Adelaide, School of Mechanical Engineering, 2007

University of Technology, Sydney. Faculty of Engineering. / The thesis focuses on issues of vehicle modelling incorporating wheel-terrain interaction and low-level control design taking into account uncertainties and input time delay. Addressing these issues is of significant importance in achieving persistent autonomy for outdoor UGVs, especially when navigating on unprepared terrains. The test-bed vehicle used for this research is retrofitted from an all-terrain 20-hp, 0.5-tonne vehicle. Its driveline system consists of an internal combustion engine, continuous variable transmission (CVT), gearbox, differential, chains, and eight wheels. The vehicle is driven in the skid-steering mode, which is popular for many off-road land-vehicle platforms. In this thesis, a comprehensive approach is proposed for modelling the driveline. The approach considers the difference in speed between two outputs of the differential and the turning mechanism of the vehicle. It describes dynamics of all components in the vehicle driveline in an integrated manner with the vehicle motion. Given a pattern of the throttle position, left and right braking efforts as the inputs, the dynamic behaviour of the wheels and other components of the UGV can be predicted. For controlling the vehicle at the low level, PID controllers are firstly used for all actuators. As many components of the vehicle exhibit nonlinearities and time delay, the large overshoots encountered in the outputs can lead to undesirable vehicle behaviours. To alleviate the problem, a novel control approach is proposed for suppression of overshoots resulting from PID control. Sliding mode control (SMC) is employed, for this, with time delay compensated by using an output predictor. As a result, the proposed approach can improve significantly system robustness and reduce substantially step response overshoot. Notably, the design is generic in that it can be applied for many dynamic processes. Knowledge of the interaction between the UGV and the terrain plays an important role in increasing its autonomy and securing the safety for off-road locomotion. In this regard, vehicle kinematic equations are combined with the theory of terramechanics for dynamic modelling of the interaction between the vehicle wheels and a variety of terrain types. Also, a fast algorithm is developed to enable online implementation. The novel interaction model takes into account the relationship between normal stresses, shear stresses, and shear displacement of the terrain that is in contact with the wheels in deriving the three-dimensional reaction forces. Finally, all modelling and control algorithms are integrated into a unique simulator for emulating the vehicle mobility characteristics. In particular, the wheel’s slip and rolling resistance can also be derived to provide useful information for closed-loop control when the UGV is navigating in an unknown environment. The simulator, as a tool for analysing the vehicle mobility, is helpful for further research on relevant topics such as traction control, safe and effective locomotion.

We study vertical jumping in a simple robot comprising an actuated mass spring arrangement. The actuator frequency and phase are systematically varied to find optimal performance. Optimal jumps occur above and below (but not at) the robot's resonant frequency f0. Two distinct jumping modes emerge: a simple jump which is optimal above f0 is achievable with a squat maneuver, and a peculiar stutter jump which is optimal below f0 is generated with a countermovement. A simple dynamical model reveals how optimal lift-off results from non-resonant transient dynamics.

In order for a mobile robot to complete its task it must be able to plan and follow a trajectory. Depending on the environment, it may also be necessary to follow a given velocity profile. This is known as temporal planning. Temporal planning can be used to minimize time of motion and to avoid moving obstacles. For example, assuming the mobile robot is an intelligent wheelchair, it must follow a prescribed path (sidewalk, hospital corridor) while following a strict speed limit (slowing down for pedestrians, cars). Computing a realistic velocity profile for a mobile robot is a challenging task due to a large number of kinematic and dynamic constraints that are involved. Unlike prior works which performed temporal planning in a 2-dimensional environment, this thesis presents a new temporal planning algorithm in a 3-dimensional environment. This algorithm is implemented on a wheeled mobile robot that is to be used in a healthcare setting. The path planning stage is accomplished by using cubic spline functions. A rudimentary trajectory is created by assigning an arbitrary time to each segment of the path. This trajectory is made feasible by applying a number of constraints and using a linear scaling technique. When a velocity profile is provided, a non-linear time scaling technique is used to fit the robots center linear velocity to the specified velocity. A method for avoiding moving obstacles is also implemented. Both simulation and experimental results for the wheeled mobile robot are presented. These results show good agreement with each other. For both simulation and experimentation, six different examples of paths in the Engineering Building of the University of Saskatchewan, were used. Experiments were performed using the PowerBot mobile robot in the robotics lab at the University of Saskatchewan.

In the future, robots will expand from industrial and research applications to the home. Domestic service robots will work in the home to perform useful tasks such as object retrieval, cleaning, organization, and security. The tireless support of these systems will not only enable able bodied people to avoid mundane chores; they will also enable the elderly to remain independent from institutional care by providing service, safety, and companionship. Robots will need to understand the relationship between objects and their environments to perform some of these tasks. Structured indoor environments are organized according to architectural guidelines and convenience for their residents. Utilizing this information makes it possible to predict the location of objects. Conversely, one can also predict the function of a room from the detection of a few objects within a given space. This thesis introduces a framework for combining object permanence and context called the probabilistic cognitive model. This framework combines reasoning about spatial extent of places and the identity of objects and their relationships to one another and to the locations where they appear. This type of reasoning takes into account the context in which objects appear to determine their identity and purpose. The probabilistic cognitive model combines a mapping system called OmniMapper with a conditional random field probabilistic model for context representation. The conditional random field models the dependencies between location and identity in a real-world domestic environment. This model is used by mobile robot systems to predict the effects of their actions during autonomous object search tasks in unknown environments.

In order for a mobile robot to complete its task it must be able to plan and follow a trajectory. Depending on the environment, it may also be necessary to follow a given velocity profile. This is known as temporal planning. Temporal planning can be used to minimize time of motion and to avoid moving obstacles. For example, assuming the mobile robot is an intelligent wheelchair, it must follow a prescribed path (sidewalk, hospital corridor) while following a strict speed limit (slowing down for pedestrians, cars). Computing a realistic velocity profile for a mobile robot is a challenging task due to a large number of kinematic and dynamic constraints that are involved. Unlike prior works which performed temporal planning in a 2-dimensional environment, this thesis presents a new temporal planning algorithm in a 3-dimensional environment. This algorithm is implemented on a wheeled mobile robot that is to be used in a healthcare setting. The path planning stage is accomplished by using cubic spline functions. A rudimentary trajectory is created by assigning an arbitrary time to each segment of the path. This trajectory is made feasible by applying a number of constraints and using a linear scaling technique. When a velocity profile is provided, a non-linear time scaling technique is used to fit the robots center linear velocity to the specified velocity. A method for avoiding moving obstacles is also implemented. Both simulation and experimental results for the wheeled mobile robot are presented. These results show good agreement with each other. For both simulation and experimentation, six different examples of paths in the Engineering Building of the University of Saskatchewan, were used. Experiments were performed using the PowerBot mobile robot in the robotics lab at the University of Saskatchewan.

Autonomous robots can replace humans to explore hostile areas, such as Mars and other inhospitable regions. A fundamental task for the autonomous robot is navigation. Due to the inherent difficulties in understanding natural objects and changing environments, navigation for unstructured environments, such as natural environments, has largely unsolved problems. However, navigation for ill-structured environments [1], where roads do not disappear completely, increases the understanding of these difficulties. We develop algorithms for robot navigation on ill-structured roads with monocular vision based on two elements: the appearance information and the geometric information. The fundamental problem of the appearance information-based navigation is road presentation. We propose a new type of road description, a vision vector space (V2-Space), which is a set of local collision-free directions in image space. We report how the V2-Space is constructed and how the V2-Space can be used to incorporate vehicle kinematic, dynamic, and time-delay constraints in motion planning. Failures occur due to the limitations of the appearance information-based navigation, such as a lack of geometric information. We expand the research to include consideration of geometric information. We present the vision-based navigation system using the geometric information. To compute depth with monocular vision, we use images obtained from different camera perspectives during robot navigation. For any given image pair, the depth error in regions close to the camera baseline can be excessively large. This degenerated region is named untrusted area, which could lead to collisions. We analyze how the untrusted areas are distributed on the road plane and predict them accordingly before the robot makes its move. We propose an algorithm to assist the robot in avoiding the untrusted area by selecting optimal locations to take frames while navigating. Experiments show that the algorithm can significantly reduce the depth error and hence reduce the risk of collisions. Although this approach is developed for monocular vision, it can be applied to multiple cameras to control the depth error. The concept of an untrusted area can be applied to 3D reconstruction with a two-view approach.

In this document, we propose and examine the novel use of a learning mechanism between the reactive and deliberative layers of a hybrid robot control architecture. Balancing the need to achieve complex goals and meet real-time constraints, many modern mobile robot navigation control systems make use of a hybrid deliberative-reactive architecture. In this paradigm, a high-level deliberative layer plans routes or actions toward a known goal, based on accumulated world knowledge. A low-level reactive layer selects motor commands based on current sensor data and the deliberative layer's plan. The desired system-level effect of this architecture is that the robot is able to combine complex reasoning toward global objectives with quick reaction to local constraints. Implicit in this type of architecture, is the assumption that both layers are using the same model of the robot's capabilities and constraints. It may happen, for example, due to differences in representation of the robot's kinematic constraints, that the deliberative layer creates a plan that the reactive layer cannot follow. This sort of conflict may cause a degradation in system-level performance, if not complete navigational deadlock. Traditionally, it has been the task of the robot designer to ensure that the layers operate in a compatible manner. However, this is a complex, empirical task. Working to improve system-level performance and navigational robustness, we propose introducing a learning mechanism between the reactive layer and the deliberative layer, allowing the deliberative layer to learn a model of the reactive layer's execution of its plans. First, we focus on detecting this inter-layer conflict, and acting based on a corrected model. This is demonstrated on a physical robotic platform in an unstructured outdoor environment. Next, we focus on learning a model to predict instances of inter-layer conflict, and planning to act with respect to this model. This is demonstrated using supervised learning in a physics-based simulation environment. Results and algorithms are presented.