In this work, a new concentric circles detection method for object detection is proposed. It has been applied to the images of a commercial radar, captured with a Charge-Coupled Device (CCD) camera. The processing includes the detection of centres and concentric circles in the images and the identification of the radar scale. Several methods found in the literature have been applied and compared with our novel proposal for multiple concentric circles detection, called “Propagation Method based on Circular Regression”. This methodology has been validated with real radar images, proving its efficiency in obtaining the distance of any object to a marine vessel, with high accuracy and low computational cost, in real time. This system can not only be applied to most existing radars in the market by adjusting the parameters of each model but our proposal for concentric circle detection can be also applied to other sensing applications.

An efficient obstacle detection technique is required so that navigating robots can avoid obstacles and potential hazards. This task is usually simplified by relying on structural patterns. However, obstacle detection constitutes a challenging problem in unstructured unknown environments, where such patterns may not exist. Talukder et al. (2002, IEEE Intelligent Vehicles Symposium, pp. 610–618.) successfully derived a method to deal with such environments. Nevertheless, the method has a high computational cost and researchers that employ it usually rely on approximations to achieve real-time. We hypothesize that by using a graphics processing unit (GPU), the computing time of the method can be significantly reduced. Throughout the implementation process, we developed a general framework for processing dynamically-sized sliding windows on a GPU. The framework can be applied to other problems that require similar computation. Experiments were performed with a stereo camera and an RGB-D sensor, where the GPU implementations were compared to multi-core and single-core CPU implementations. The results show a significant gain in the computational performance, i.e. in a particular instance, a GPU implementation is almost 90 times faster than a single-core one.

We investigated the design parameters of a compact pot-like ultrasonic sensor which possesses a highly anisotropic beam pattern. As the sensor size is small due to its application constraint, the parameters are thus highly coupled to one another. We analyzed the respective effects of the parameters in the case where there is a vertical beam width reduction. The parameters investigated include resonant frequency, vibrating plate width-expanded angle, and ratio of thickness discontinuity of the vibrating plate. Numerical models developed by combining finite-element analysis and spatial Fourier transforms were adopted to predict the far-field radiating beam pattern of the various design configurations. The displacement distribution of the vibrating plate was measured using a microscopic laser Doppler vibrometer and the far-field pressure beam patterns were measured using a standard microphone in a semianechoic environment. The three configurations we used to validate the simulation model resulted in an H-V ratio of 2.67, 2.68 and 3.13, respectively which all agreed well with the numerical calculations. We found that by increasing the operating resonant frequency from 40kHz to 58kHz, the vertical far-field beam width of an ultrasonic sensor can be reduced by 31.62%. We found that the vertical beam width can be significantly reduced when the ratio of the thickness discontinuity of the vibrating plate decreases from 1 to 0.4 and is incorporated with its optimal width-expanded angle of the vibrating plate. It appears that an ultrasonic sensor with this type of anisotropic beam pattern can be ideally adopted for today's automotive applications.

This paper presents a computationally simple stereo vision method for detecting useful features of obstacles on the ground relevant to a mobile robot's navigation in an indoor environment. Enhanced time efficiency and reliability is achieved by introducing a geometrical image transformation and a frame-wise iterative edge image comparison scheme. The image transformation used in this paper relates each constant disparity value to an oblique plane at an elevation from the floor. The stereo correspondence method is devised by implementing a frame-wise AND operation based upon edge orientation and a subsequent pixel wise intensity correlation checking. This method is efficient in terms of computation time and capable of identifying isodisparity points, so that the height information of all the obstacle points above the floor can be determined. The obtained disparities are recorded on a local map to give complete obstacle features for mobile robot's path planning. Through a series of experiments performed under various environmental conditions, it is found that the proposed method can effectively be applied for locating obstacles of various heights in indoor navigation of mobile robots.