Praneel Chand

Mapping and exploration in a hierarchical heterogeneous multi-robot system using limited capability robots

This paper focusses on the development of a customised mapping and exploration task for a heterogeneous ensemble of mobile robots. Many robots in the team may have limited processing and sensing abilities. This means that each robot may not be able to execute all components of the mapping and exploration task. A hierarchical system is proposed that consists of computationally powerful robots (managers) at the upper level and limited capability robots (workers) at the lower levels. This enables resources (such as processing) to be shared and tasks to be abstracted. The global environment containing scattered obstacles is divided into local environments by the managers for the workers to explore. Worker robots can be assigned planner and/or explorer tasks and are only made aware of information relevant to their assigned tasks.

A two-tiered global path planning strategy for limited memory mobile robots

Multi-robot systems have inherent advantages such as the ability to allocate and redistribute tasks across the team of robots. For multi-robot tasks such as exploration of large environments, some of the available robots may only possess simple embedded controllers with limited memory capacity. However, in some situations these limited robots may be required to perform global path planning to navigate beyond localised regions of the large environment. Global path planning can be problematic for the limited memory robots if they are unable to store the entire map in their local memory. Hence, this paper presents and evaluates a two-tiered path planning technique to permit global path planning. A set of local maps describing the global map is searched using a two-tiered A^* algorithm that executes entirely on the limited memory robots. Planning time, data communication and path length are evaluated for various combinations of local and global maps. Employing smaller local map sizes in large global maps is capable of yielding superior or comparable execution times to non-memory constrained planning.

Object detection and recognition for a pick and place Robot

Controlling a Robotic arm for applications such as object sorting with the use of vision sensors would need a robust image processing algorithm to recognize and detect the target object. This paper is directed towards the development of the image processing algorithm which is a pre-requisite for the full operation of a pick and place Robotic arm intended for object sorting task For this type of task first the objects are detected, and this is accomplished by feature extraction algorithm. Next, the extracted image (parameters in compliance with the classifier) is sent to the classifier to recognize what object it is and once this is finalized, the output would be the type of the object along with its coordinates to be ready for the Robotic Arm to execute the pick and place task The major challenge faced in developing this image processing algorithm was that upon making the test subjects in compliance with the classifier parameters, resizing of the images conceded in the loss of pixel data. Therefore, a centered image approach was taken. The accuracy of the classifier developed in this paper was 99.33% and for the feature extraction algorithm, the accuracy was 83.6443%. Finally, the overall system performance of the image processing algorithm developed after experimentation was 82.7162%.

Development of a navigation system for heterogeneous mobile robots

Navigation systems often play an important role in mobile robot control. Many existing mobile robot navigation systems have been implemented and tested for particular types of robots. However, in some implementations, such as heterogeneous multi-robot systems, a generic navigation system can offer potential advantages. In such applications, a generic navigation system should be able account for robots with varying size, shape, drive type and sensor quantities. Additionally, it should be capable of offering a high degree of flexibility for navigation in known and unknown environments. Hence, a single generic navigation system that combines the benefits of reactive and deliberative control has been developed for heterogeneous mobile robots. The design of the hierarchical hybrid navigation system is based on the A* algorithm, a polar histogram and a modified dynamic window approach. Simulation experiments with three heterogeneous robots in a range of environments have been conducted. Performance of reactive navigation and hybrid reactive-deliberative navigation in known and unknown environments is evaluated. Favourable results are achieved with the developed reactive system. Hybrid reactive-deliberative navigation offers improved performance over reactive navigation in known environments. Deliberative control does not affect performance significantly in unknown environments. Initial hardware experiments demonstrate that the navigation system can work on real robots.

Feedback Coordination of Limited Capability Mobile Robots

In some multi-robot applications, such as exploration, predefined task allocation and coordination can fail to function adequately. This failure is attributed to the inability to completely model a robots interactions with the environment before task execution. Hence, the main focus of this paper is on a feedback coordination mechanism that executes periodically after initial task allocation. This feedback mechanism monitors the individual and group performance of worker robots. If the performance of a worker robot is unsatisfactory, a task reallocation algorithm adjusts the task-robot combinations of the team. Three cases of unsatisfactory robot performance that can be detected by the feedback mechanism include: complete failure, partial failure, and poor performance. The task allocation and coordination strategy is applied to a multi-robot exploration task. Initial results from experiments in various types of environments indicate that the feedback coordination mechanism successfully identifies the three forms of robot failure, and improves the systems performance.

The Design of a Pair of Identical Mobile Robots to Investigate Cooperative Behaviours

mechanical size and weight of devices being built, with a proportional increase in expense. An alternative to this conventional trend is the use of cooperative behaviour. Humans are limited by their physical size and strength – and yet are capable of moving heavy furniture and other large items outside their individual strength ability. This is accomplished with the help of another person to share the load, or a large number of people for a substantial object. Requirements of coordination, communication and precision control have restricted this approach being considered for mobile robots in the past. However, more powerful processors and software are removing these barriers, permitting cooperative robots to be seriously considered for a number of applications. By working in parallel, cooperative behaviour can increase efficiency and reduce the time required to complete a task. Reliability is increased by introducing redundancy when using a team of robots, while cost is reduced due to the use of smaller simplistic machine designs. Application specific design and manufacturing costs can be removed by fabricating the robots semi-generically. Reduced weight means less preparation and upkeep of the working environment, and new complex tasks can be introduced which are too difficult for a single entity to achieve. We demonstrate a methodology for the construction of a low-cost, versatile, highly manoeuvrable and computationally powerful robot capable of autonomous operation. We demonstrate that these robots can be fabric ated at a small fraction of the price of equivalent commercial systems. As our interest is in the development of cooperative robotic behaviour, the software control system, machine interface and an introduction to cooperative robotic systems are also discussed.

Inverse kinematics solution for trajectory tracking using artificial neural networks for SCORBOT ER-4u

This paper presents the kinematic analysis of the SCORBOT-ER 4u robot arm using a Multi-Layered Feed-Forward (MLFF) Neural Network. The SCORBOT-ER 4u is a 5-DOF vertical articulated educational robot with revolute joints. The Denavit-Hartenberg and Geometrical methods are the forward kinematic algorithms used to generate data and train the neural network. The learning of forward-inverse mapping enables the inverse kinematic solution to be found. The algorithm is tested on hardware (SCORBOT-ER 4u) and reliable results are obtained. The modeling and simulations are done using MATLAB 8.0 software.

Design and development of low cost voice control smart home device in the South Pacific

To accomplish control tasks, an emerging innovative method of automation is voice control. In conjunction with Lab VIEW [1] and Windows Speech recognition [2], the two most human interactive activities; drawing of curtains and switching on and off of lights are controlled. This will help in improving living standard by making living easier for elderly people, reducing frustration and increasing independence at senior homes in the South pacific. The price comparison indicates that 81.25% saving can be made with this product in regards to energy consumption.

Development of a reduced human user input task allocation method for multiple robots

Task allocation mechanisms are employed by multi-robot systems to efficiently distribute tasks between different robots. Currently, many task allocation methods rely on detailed expert knowledge to coordinate robots. However, it may not be feasible to dedicate an expert human user to a multi-robot system. Hence, a non-expert user may have to specify tasks to a team of robots in some situations. This paper presents a novel reduced human user input multi-robot task allocation technique that utilises Fuzzy Inference Systems (FISs). A two-stage primary and secondary task allocation process is employed to select a team of robots comprising manager and worker robots. A multi-robot mapping and exploration task is utilised as a model task to evaluate the task allocation process. Experiments show that primary task allocation is able to successfully identify and select manager robots. Similarly, secondary task allocation successfully identifies and selects worker robots. Both task allocation processes are also robust to parameter variation permitting intuitive selection of parameter values.

Towards a robust feedback system for coordinating a hierarchical multi-robot system

Restricting the usage of a team of robots to a few expert human users can be disadvantageous. In applications such as exploration, it may not always be possible for human experts to travel to sites, resulting in negative consequences. It is preferable to have a robotic system that is capable of coordinating itself based on inputs provided by non-expert human users. Hence, this paper presents the development of a robust feedback system for coordinating a hierarchical team of robots where inputs are specified by non-expert human users. Experiments with a multi-robot mapping and exploration task show that the feedback system successfully detects and corrects three types of failures. These are poor performance, partial failure and complete failure. Moreover, the system is robust to threshold value variation and monitor time interval variation within the tested limits.