Search

Image based Localization (IbL) uses both Structure from Motion (SfM) and Simultaneous Localization and Mapping (SLAM) data for accurate pose estimation. However, under conditions where there is a large perspective difference between the SfM images and SLAM keyframes, the SfM-SLAM co-visibility graph becomes sparse. As a result, the scale drift can increase especially when using monocular SLAM as part of the IbL framework. The drift rarely gets corrected at loop closure due to its large magnitude. We propose a split affine transformation approach that uses SfM-SLAM information along with Sim(3) optimization to minimize the scale drift. Experiments are performed using an image dataset collected in a campus environment with different trajectories, showing the improvement in scale drift correction with the proposed method. The SLAM data was collected close to plainly textured structures like buildings while SfM images were captured from a larger distance from the building facade which leads to a challenging navigation scenario in the context of IbL. Localizing mobile platforms moving close to buildings is an example of such a case. The paper positively impacts the widespread use of small autonomous robotic platforms, which is to perform an accurate outdoor localization under urban conditions using only a monocular camera.

The movement of cooperative robots in a densely cluttered environment may be impossible if the formation type is invariant. Hence, we investigate a novel method of switching time-invariant formation control for a group of heterogeneous autonomous vehicles, which may include Unmanned Ground Vehicles (UGV) and Unmanned Aerial Vehicles (UAV). We have extended a Negative Imaginary (NI) consensus control approach to switch the formation shape of the robots whilst only using the relative distances between agents, in combination with those between agents and obstacles. All agents can automatically create a new safe formation to overcome obstacles, then restore the prototype formation once the obstacles are cleared. We address the oscillation and local minima problems, generated by existing algorithms, using a newly developed circle fitting approach. We also highlight the advantages of our proposed algorithms over the kinematic controller and the conventional Artificial Potential Field (APF) controller from both theoretical and experimental standpoints. Simulation and experimental results are then analyzed to validate the feasibility of our proposed approach.

This paper presents a hybrid non-linear control system, comprised of a conventional PD controller and a PD-type fuzzy logic autopilot for high performance trajectory tracking of a quadcopter drone. Given the nature of the traditional control approaches, which are model-based; and the fuzzy logic control systems, which are knowledge-based; the proposed hybrid controllers can provide a more robust solution in the face of uncertainties. Both controllers operate in a parallel incremental form to improve the transient performance and the robustness of the closed loop control system. Through extensive computer simulations and real-time flight tests, this paper highlights the efficacy of the proposed control system in the face of parameter variations, non-liner aerodynamic models, and external disturbances (e.g. wind gusts). The Dryden and 1-cos turbulence models are employed to represent wind gusts under realistic flight environments. Stability analysis of the closed loop control system is conducted using the Lyapunov's indirect method.

We introduce a new non-linear system identification technique, leveraging on the benefits of the Evolutionary Takagi-Sugeno fuzzy system and the Type-2 interval fuzzy systems. Our technique has the ability to learn-from-scratch while having a better ability to accommodate the footprint-of-uncertainties (FoU). To support its mission in meeting the challenge in uncertain non-linear dynamic systems, we also introduce a new type reduction method to convert the Type-2 fuzzy systems into its Type-1 counterparts. As a part of its pruning strategy, the proposed system incorporates the concept of information entropy to avoid over-fitting, which is a highly undesirable issue in modeling. We demonstrate the effectiveness of our technique in achieving a delicate balance between minimizing the complexity of the structure of the acquired fuzzy model while maximizing the prediction accuracy. We employ a set of challenging pH neutralization data, known for its substantial non-linearity and large variation, in addition to a non-linear mechanical system. We conclude our research by conducting a rigorous comparative study to quantify the relative merits of our proposed technique with respect to the previous ETS algorithm (as its predecessor), the well-known KM-type reduction algorithm, and the higher-order discrete transfer functions, widely implemented in conventional mathematical modeling techniques.

Considerable progress has been made in improvingthe estimation accuracy of cognitive workload using varioussensor technologies. However, the overall performance of different algorithms and methods remain sub-optimal in real-worldapplications. Some studies in the literature demonstrate thata single modality is sufficient to estimate cognitive workload.These studies are limited to controlled settings, a scenario thatis significantly different from the real-world where data getscorrupted, interrupted, and delayed. In such situations, the useof multiple modalities is neededMulti-modal fusion approaches have been successful in otherdomains, such as wireless sensor networks, in addressing singlesensor weaknesses and improving information quality/accuracy.These approaches are inherently more reliable when a datasource is lost. In the cognitive workload literature, sensors suchas EEG, ECG, and eye tracking have shown success in estimatingaspects of cognitive workload. Multi-modal approaches thatcombine data from several sensors together can be more robustfor real-time measurement of cognitive workload. In this paper,we review published studies related to multi-modal data fusion forestimating cognitive workload and synthesise their main findings.We identify opportunities for designing better

The development of Unmanned Aerial Vehicles (UAVs) has become one of the most fruitful research areas in the field of autonomous flight ontrol. Quadcopters are chosen due to their simple mechanical structure which is able to hover in a stationary manner, vertical take-off and landing Nevertheless, these types of aircraft are highly nonlinear and under actuated systems. Intelligent control such as fuzzy logic is a suitable choice for controlling nonlinear systems. However, type-1 fuzzy control cannot handle uncertainties of nonlinear systems. As a solution, interval type-2 fuzzy control has the advantage of being able to deal with uncertainties. This research proposes the use of interval type-2 fuzzy for controlling the altitude of a nonlinear quadcopter UAV using Gaussian membership function and utilizing the Enhance Iterative Algorithm with Stop Condition (EIASC) algorithm for type-reduction. A comparison between the proposed interval type-2 fuzzy controller and proportional-Derivative (PD) controller is illustrated. Simulation results demonstrated that the tracking performance of the proposed controller outperformed the PD controller.

In this paper, an evolutionary general regression neural network is developed based on limited incremental evolution and distance-based pruning to online model dynamic systems. Also, a variance-based method is suggested to adapt the smoothing parameter in GRNN for online applications. The proposed model is compared with different types of dynamic neural networks. A nonlinear benchmarking dynamic discrete system with white Gaussian noise is used in the comparison. The results are compared in terms of the prediction error and the time required for adaption and the comparison results show that the proposed model is more accurate and quicker than any another counterpart.

The public debate and discussion about trust in Computational Intelligence (CI) systems is not new, but a topic that has seen a recent rise. This is mainly due to the explosion of technological innovations that have been brought to the attention of the public, from lab to reality usually through media reporting. This growth in the public attention was further compounded by the 2018 GDPR legislation and new laws regarding the right to explainable systems, such as the use of “accurate data”, “clear logic” and the “use of appropriate mathematical and statistical procedures for profiling”. Therefore, trust is not just a topic for debate – it must be addressed from the onset, through the selection of fundamental machine learning processes that are used to create models embedded within autonomous decision-making systems, to the selection of training, validation and testing data. This paper presents current work on trust in the field of Computational Intelligence systems and discusses the legal framework we should ascribe to trust in CI systems. A case study examining current public perceptions of recent CI inspired technologies which took part at a national science festival is presented with some surprising results. Finally, we look at current research underway that is aiming to increase trust in Computational Intelligent systems and we identify a clear educational gap.

Human land use threatens global biodiversity and compromises multiple ecosystem functions critical to food production. Whether crop yield-related ecosystem services can be maintained by a few dominant species or rely on high richness remains unclear. Using a global database from 89 studies (with 1475 locations), we partition the relative importance of species richness, abundance, and dominance for pollination; biological pest control; and final yields in the context of ongoing land-use change. Pollinator and enemy richness directly supported ecosystem services in addition to and independent of abundance and dominance. Up to 50% of the negative effects of landscape simplification on ecosystem services was due to richness losses of service-providing organisms, with negative consequences for crop yields. Maintaining the biodiversity of ecosystem service providers is therefore vital to sustain the flow of key agroecosystem benefits to society.

In this research, we develop a self-learning fuzzy logic autopilot for high-performance trajectory tracking of a hexacopter unmanned aerial vehicle. We employ non-linear mathematical models of the system, derived from first principles, to gain more accurate understanding of its dynamic behaviours. Accordingly, we design the TS-fuzzy autopilot of a hexacopter for its altitude loop using the C-means fuzzy clustering technique to control the height of the drone. This research serves as our preliminary study to investigate the feasibility of our fuzzy control system before we can implement it into practice. We perform a systematic comparative study to highlight the effectiveness of our control algorithm. We demonstrate the performance and robustness of the proposed control system in terms of its tracking error with respect to the performance of the conventional PID controller.

In this work we present non-linear system identification for a class of the dragonfly-like flapping wing aircraft. We model the system in its vertical and all attitude loops (roll, pitch, and yaw) as well as its actuator dynamics. Based on a set of input-output data, obtained from first principle modelling; we perform the entropy fuzzy system identification to derive the open loop dynamics of the aircraft using the Mamdani Fuzzy inference method, which is more intuitive, despite being non-linear. Thiswill make the proposed models well-suited to non-expert users(e.g. average drone operators). Our research indicates that theinformation entropy is very effective to maximize the system accuracy while avoiding overfitting problems. Through numerical simulation, we demonstrate the efficacy of the proposed fuzzy models as we can achieve reasonably good average modelling accuracy of around 90 % for all attitude loops.

This paper deals with an approach to path planning by obstacle avoidancefor multi-agent systems. An effective framework is presented based on theParticle Swarm Optimization (PSO) method; an evolutionary computation(EC) technique that uses the dynamics of the swarm to search the solutions for theoptimization problems. It describes the path replanning technique and obstacleavoidance for autonomous multi-agent systems. A simultaneous replanning conceptis incorporated into the path planning to avoid both static and dynamic obstacles.This proposed algorithm reduces the computational time of the path planning. In thedynamic environment, the numerical results show that the Simultaneous ReplanningVectorized Particle Swarm Optimization (SRVPSO) algorithm is effective andalso efficient for multi-agent systems.

© 2015 IEEE.The paper shows the novel way of detecting the obstacles, so that the intelligence of the autonomous vehicle (AV) can be enhanced for the path planning module. Path planning in cluttered environment has been considered as the main challenge in the field of autonomous vehicles. These vehicles need to be intelligent enough to recognize their surroundings and avoid the obstacles effectively. In cluttered environment, the AV requires the comprehensive perception of the environment to avoid the obstacles. So it is better to know about the kinematic behaviour of the obstacles while planning their path. Hence, the orientation of moving obstacle is estimated based on comparing the 3D range images between successive time frames. These 3D range information are acquired from a single vision sensor. The paper uses the differential scene flow technique to extract the flow vectors in 3D coordinates. Gradient Vector Field (GVF) method is utilized effectively in scene flow to extract changes in pixel values in the three directions which are used to recognize the obstacles. Our aim is to develop an autonomous path planning algorithm with closed-loop motion control based on scene flow which rely on a single vision sensor and an on-board hardware. Experimental results are discussed finally.

Sexual reproduction in animals requires close interactions with the opposite sex. These interactions may generate costs of reproduction, because mates can induce detrimental physiological or physical effects on one another, due to their interest in maximising their own fitness. To understand how a male's presence influences aspects of female physiology implicated in reproductive costs in mice, independent of offspring production, we paired females with vasectomised, castrated or intact males, or other females. Being paired with a male, irrespective of his gonadal status, increased female weight. This effect was transient in females paired with castrated males but more persistent in those with vasectomised males. Those paired with males also showed an increase in corticosterone, suggesting an increased stress response. However, this was dependent on the gonadal status of the male housing partner, since those housed with vasectomised males had lower corticosterone than those with castrated males. Altered energy metabolism was only detectable in pregnant females, and oxidative stress was not consistently affected by a female's housing partner. These results suggest that a male's presence alters female weight, and stresses associated with reproduction could be induced by simply the presence of a male, but reduced by mating and/or being solicited to mate.