Andrew Bainbridge-Smith

Altitude control of a quadrotor helicopter using depth map from Microsoft Kinect sensor

Reliable depth estimation is a cornerstone of many autonomous robotic control systems. The Microsoft Kinect is a new, low cost, commodity game controller peripheral that calculates a depth map of the environment with good accuracy and high rate. In this paper we calibrate the Kinect depth and image sensors and then use the depth map to control the altitude of a quadrotor helicopter. This paper presents the first results of using this sensor in a real-time robotics control application.

Determining optical flow using a differential method

Abstract A comparison of different methods for determining optical flow based on local information is presented. It is shown that to overcome the aperture problem, second order differential terms of the image function are required. It is also argued that the fundamental problem of measuring optical flow is to determine the velocity in as small an aperture as possible. It is concluded that the technique of Lucas and Kanade (see Barron et al., Performance of optical flow techniques, Int. J. Computer Vision , 12 (1) (1994) 43–77) is the best generalised second order method for achieving this objective.

Multiply Accumulate Unit Optimised for Fast Dot-Product Evaluation

A fast dot-product unit suitable for long word lengths is shown. Its implementation is based on computing only the significant partial products and exploiting the properties of the asymmetric signed digit redundant number representation. Optimal partial product packing and a carry propagation free adder combine to yield a MAC with high throughput. An example design of a low-pass FIR filter of 51 taps of 32 bit word-length was synthesised for the Altera cyclone II FPGA family. A filter clock speed of 220 MHz and a throughput of 12.9 MSamples/s was achieved.

A Common Subexpression Elimination Tree Algorithm

A common subexpression elimination algorithm is proposed to minimize the complexity of the multiple constant multiplication operation. The coefficients (constants) of the multiple constant multiplication are represented using the binary signed digit number system. The binary signed digit representations of each coefficient are enumerated using the representation tree. The algorithm traverses the tree to calculate the possible subexpressions at each node. Each subexpression is used to find a possible decomposition for the coefficient to be encoded. A complexity formula is proposed to compare the decompositions. The algorithm is designed to prune the tree when it finds a decomposition with minimum complexity. This reduces the search space while minimizing the hardware complexity. Results show that the algorithm has better performance than other published algorithms including linear programming optimization methods. The algorithm outperforms the subexpression sharing method in that uses only the canonical signed-digit representations.

Biologically inspired UAV obstacle avoidance and control using monocular optical flow & divergence templates

This paper describes a biologically inspired flight control strategy for unpiloted aerial vehicles (UAVs) using optical flow and depth information. The scene depth map is generated by measurement of optical flow in a uniform grid across the image. Grid regions are each analyzed singularly, each ones depth is estimated by comparing their observed motion with the motion of the craft as measured by the inertial measurement unit. Two control processes run simultaneously on the robot, a gross strategy uses divergence of optical flow vectors about the focus of expansion, balancing these according to a predefined divergence template. A fine strategy looks for outliers from the estimated depth map. The sum of these processes generates a control impulse to steer a quadrotor helicopter away from obstacles. The algorithm was tested against ground-truth datasets and real flight. In both situations it was able to control the craft heading and pitch during indoor operation, avoiding both corridor walls and oncoming obstacles.

ADH: an aspect described hardware programming language

This paper describes preliminary work undertaken in the development of a new hardware description programming language (ADH). ADH is intended for the signal processing domain and is designed around aspect oriented programming rather than procedural or object models.

Efficient implementation of fast redundant number adders for long word-lengths in FPGAs

The use of redundant number systems can significantly improve computational performance in numerically intensive applications, however, the implementation of their arithmetic circuits is usually expensive because multiple bits are needed for each symbol (digit). This paper presents efficient adder circuits specifically targeted to the low cost FPGA architectures of the Xilinx Spartan 3 and the Altera Cyclone III. The special carry logic and fast carry chains are re-purposed to serve the new adders. These circuits use the redundancy in representation to eliminate carry propagation, providing near constant addition delay irrespective of the operand width. This is confirmed experimentally and shown to outperform the architecture optimised binary ripple carry adders. The critical path delay cross over for the binary and binary signed digit adders are at widths of 44 and 24 symbols, using only 2 and 3 times the number of look-up tables on the Spartan 3 and Cyclone III respectively. Fast prefix-tree adders do not compare favourably at any width.

Optical Flow for Heading Estimation of a Quadrotor Helicopter

This paper demonstrates that computer vision techniques can estimate the heading of a small fixed pitch four rotor helicopter. Heading estimates are computed using the optical flow technique of phase correlation on images captured using a down facing camera. The camera is fitted with an omnidirectional lens and the images are transformed into the log-polar domain before the main computational step. The vision algorithm runs at 10 Hz on a single board computer (SBC) mounted aboard the craft. Experimental performance of this system is compared with results obtained from a traditional inertial measurement unit (IMU). It is found that the yaw rate computed from the optical flow is comparable to the IMU and thus appropriate for use in controlling the helicopter.

Determination of myosin filament positions and orientations in electron micrographs of muscle cross sections

An automated image analysis system for determination of myosin filament orientations in electron micrographs of muscle cross-sections is described. Analysis of the distribution of the orientations is important in studies of muscle structure, particularly for interpretation of x-ray diffraction data. Filament positions are determined using h-dome extraction and image filtering, based on grayscale reconstruction. Erroneous locations are eliminated based on lattice regularity. Filament orientations are determined by correlation with a template that incorporates the salient filament characteristics and classified using a Gaussian mixture model. Application to a number of micrographs and comparison with manual classifications of orientations shows that the system is effective in many cases.

Beyond optical flow — Biomimetic UAV altitude control using horizontal edge information

Biologically inspired unpiloted aerial vehicle (UAV) flight control typically uses regulation of optical flow magnitude or distribution. These designs were motivated by research undertaken on insects that showed they too used optical flow in their flight. Recent research has suggested that optical flow alone can not sufficiently explain the altitude control behaviour of honeybees and houseflies; those insects must consider other environmental features like local edges. Inspired by these results, this paper presents a biomimetic altitude control and avoidance strategy which analyzes horizontal edges in the image and is able to successfully control the flight of a quadrotor helicopter in a simulated environment.