Insects use an estimate of the angular speed of the visual image across the eye (termed optic flow) for a wide variety of behaviors including flight speed control, visual navigation, depth estimation, grazing landings, and visual odometry. Despite the behavioral importance of visual speed estimation, the neuronal mechanisms by which the brain extracts optic flow information from the retinal image remain unknown. This dissertation investigates the underlying neuronal mechanisms of visual speed estimation via three complementary strategies: the development of neuronally-based computational models, testing of the models in a behavioral simulation framework, and behavioral experiments using bumblebees. Using these methods I demonstrate the sufficiency of two non-directional models of motion detection for reproducing real-world, speed dependent behaviors, propose potential neuronal circuits by which these models may be physiologically implemented, and predict the expected responses of these neurons to a range of visual stimuli.

No description available.

This thesis makes a number of new contributions to control and sensing for unmanned vehicles. I begin by developing a non-linear simulation of a small unmanned helicopter and then proceed to develop new algorithms for control and sensing using the simulation. The work is field-tested in successful flight trials of biologically inspired vision and neural network control for an unstable rotorcraft. The techniques are more robust and more easily implemented on a small flying vehicle than previously attempted methods.¶ Experiments from biology suggest that the sensing of image motion or optic flow in insects provides a means of determining the range to obstacles and terrain. This biologically inspired approach is applied to control of height in a helicopter, leading to the World’s first optic flow based terrain following controller for an unmanned helicopter in forward flight. Another novel optic flow based controller is developed for the control of velocity in hover. Using the measurements of height from other sensors, optic flow is used to provide a measure of the helicopters lateral and longitudinal velocities relative to the ground plane. Feedback of these velocity measurements enables automated hover with a drift of only a few cm per second, which is sufficient to allow a helicopter to land autonomously in gusty conditions with no absolute measurement of position.¶ New techniques for sensor fusion using Extended Kalman Filtering are developed to estimate attitude and velocity from noisy inertial sensors and optic flow measurements. However, such control and sensor fusion techniques can be computationally intensive, rendering them difficult or impossible to implement on a small unmanned vehicle due to limitations on computing resources. Since neural networks can perform these functions with minimal computing hardware, a new technique of control using neural networks is presented. First a hybrid plant model consisting of exactly known dynamics is combined with a black-box representation of the unknown dynamics. Simulated trajectories are then calculated for the plant using an optimal controller. Finally, a neural network is trained to mimic the optimal controller. Flight test results of control of the heave dynamics of a helicopter confirm the neural network controller’s ability to operate in high disturbance conditions and suggest that the neural network outperforms a PD controller. Sensor fusion and control of the lateral and longitudinal dynamics of the helicopter are also shown to be easily achieved using computationally modest neural networks.

Vision provides the primary means by which many animals distinguish foreground objects from their background and coordinate locomotion through complex environments. The present thesis focuses on mechanisms within the visual system that afford figure-ground segregation and self-motion perception. These processes are modeled as emergent outcomes of dynamical interactions among neural populations in several brain areas. This dissertation specifies and simulates how border-ownership signals emerge in cortex, and how the medial superior temporal area (MSTd) represents path of travel and heading, in the presence of independently moving objects (IMOs). Neurons in visual cortex that signal border-ownership, the perception that a border belongs to a figure and not its background, have been identified but the underlying mechanisms have been unclear. A model is presented that demonstrates that inter-areal interactions across model visual areas V1-V2-V4 afford border-ownership signals similar to those reported in electrophysiology for visual displays containing figures defined by luminance contrast. Competition between model neurons with different receptive field sizes is crucial for reconciling the occlusion of one object by another. The model is extended to determine border-ownership when object borders are kinetically-defined, and to detect the location and size of shapes, despite the curvature of their boundary contours. Navigation in the real world requires humans to travel along curved paths. Many perceptual models have been proposed that focus on heading, which specifies the direction of travel along straight paths, but not on path curvature. In primates, MSTd has been implicated in heading perception. A model of V1, medial temporal area (MT), and MSTd is developed herein that demonstrates how MSTd neurons can simultaneously encode path curvature and heading. Human judgments of heading are accurate in rigid environments, but are biased in the presence of IMOs. The model presented here explains the bias through recurrent connectivity in MSTd and avoids the use of differential motion detectors which, although used in existing models to discount the motion of an IMO relative to its background, is not biologically plausible. Reported modulation of the MSTd population due to attention is explained through competitive dynamics between subpopulations responding to bottom-up and top- down signals.

京都大学 / 0048 / 新制・課程博士 / 博士(文学) / 甲第20828号 / 文博第758号 / 新制||文||655(附属図書館) / 京都大学大学院文学研究科行動文化学専攻 / (主査)教授 蘆田 宏, 教授 板倉 昭二, 教授 Anderson James Russell, 准教授 ALTMANN Christian / 学位規則第4条第1項該当 / Doctor of Letters / Kyoto University / DFAM

There is considerable evidence suggesting an innate linkage between the human perceptual and motor systems, which evolve together and assist one another in the production and coordination of movement. A major contributor to this relationship is optic flow, providing movement variables such as navigation, obstacle avoidance, and depth perception. The absence of optic flow leads to the decoupling of perception and action, which has been shown to contribute to decrements in human movement (i.e., negatively impacted locomotion and posture, and slower adaptation to gait perturbations). Despite the importance of maintaining this linkage, optic flow manipulations are often found to be underrepresented in locomotion literature when specifically related to rehabilitation training (i.e., treadmills). This may be a contributor to the lengthy and exhaustive treatment plans. The literature has shown instances where reintroducing optic flow into training protocols has shown larger gait improvements in shorter times than typical ambulation protocols, however, the strength of the perception-action linkage in adulthood is still not well understood and its impact not yet fully explored. Therefore, the current research program aims to fill this gap by evaluating how the reintroduction of optic flow into atypical gait training protocols in both healthy and gait-compromised individuals may provide evidence that could be used to enhance rehabilitative outcomes. This series of conceptually related experiments explores outcome enhancements through neuromuscular level changes (Study One), the recalibration process of perception-action given newly acquired physical constraints (Study Two), and on larger scale gait cycle performances in a rehabilitation setting (Study Three). It is hypothesized that perception- action coupling will lead to increases in neuromuscular elicitation in the absence of voluntary movement (Study One), assist the recalibration process to improve measures of spatial awareness and atypical gait parameters (Study Two), and finally, improve rehabilitative outcomes in a spinal cord injury (SCI) ambulation protocol, both objectively (i.e., gait parameters, dynamic balance, SCI measures) and subjectively (i.e., questionnaires) (Study Three). / Thesis / Master of Science (MSc)

No description available.

This paper describes a complete algorithm to estimate the motion of a fixed-wing aircraft given a series of digitized flight images. The algorithm was designed for fixed-wing aircraft because carefully procured flight images and corresponding navigation data were available to us for testing. After image pre-processing, optic flow data is determined by automatically finding and tracking good features between pairs of images. The image coordinates of matched features are then processed by a rigid-object linear optic flow-motion estimation algorithm. Input factors are weighed to provide good testing techniques. Error analysis is performed with simulation data keeping these factors in mind to determine the effectiveness of the optic flow algorithm. The output of this program is an estimate of rotation and translation of the imaged environment in relation to the camera, and thereby the airplane. Real flight images from NASA test flights are used to confirm the accuracy of the algorithm. Where possible, the estimated motion parameters are compared with recorded flight instrument data to confirm the correctness of the algorithm. Results show that the algorithm is accurate to within a degree provided that enough optic flow feature points are tracked. / Master of Science

De nos jours, l'engouement pour la robotique autonome ne cesse d'augmenter en particulier pour les microdrones. En effet, ces aéronefs de petite taille font l'objet de nombreuses recherches afin de les miniaturiser et de rendre leur navigation plus autonome. Ainsi, cette thèse explore un système de vision parcimonieux dédié à la navigation courte portée au moyen de capteurs visuels auto-adaptatifs innovants composés de seulement 12 pixels aux propriétés optiques inspirées de celles de l'abeille. Deux algorithmes de mesure de flux optique sont ensuite comparés en conditions idéales sur 5 décades d'irradiance et 3 décades de vitesses optiques, puis testés en conditions réelles de vol. L'algorithme le plus robuste et le plus efficace, de par ses très faibles besoins calculatoires, a été embarqué à bord d'un micro quadrirotor pesant environ 400 g et équipé d'un système visuel parcimonieux de 96 pixels stabilisé via une nacelle articulée en roulis et tangage pour compenser les rotations du quadrirotor. Les stratégies de navigation observées chez l'abeille ont ensuite été simulées dans des environnements virtuels (tunnel de longueur 6 m ou 12 m pour une section minimale de 25 ou 50 cm) et la preuve de faisabilité de la détection du flux optique à bord d'un microdrone a été démontrée en conditions réelles de vol en salle expérimentale (vol de 4 m de long à une distance minimale de 50 cm). Couplé à des stratégies de navigation inspirées de l’abeille, ce système visuel innovant dédié à la perception du mouvement permettra dans un futur proche de naviguer dans des environnements encombrés ou exigus. / The interest in autonomous robotics is continually expanding, especially in the domain of micro air vehicles. Indeed, much research focuses on these small-size aircraft in order to miniaturize them and to make their navigation more autonomous. This PhD thesis explores a parsimonious vision system dedicated to short range navigation using innovative self-adaptive visual sensors composed of only 12 pixels with optical properties inspired by those of honeybees. Two optic flow measurement algorithms are first compared under ideal conditions over 5 decades of irradiance and 3 decades of optical velocity, then tested under real flight conditions. The most robust and efficient algorithm, due to its very low computing requirements, was embedded on board a micro quadrotor weighing about 400 g and equipped with a parsimonious visual system of 96 pixels stabilized via an articulated gimbal in roll and pitch to compensate the quadrotor rotations. The navigation strategies observed in honeybees were simulated in virtual environments (6 m or 12 m long tunnel for a minimum section of 25 or 50 cm) and the feasibility of the detection of the optic flow on board a micro quadrotor was demonstrated in real flight conditions in experimental room (flight of 4 m long at a minimum distance of 50 cm). Coupled with navigation strategies inspired by the honeybee, this innovative visual system dedicated to the perception of movement will in the near future allow to navigate in cluttered or cramped environments.

Orientador: Lilian Teresa Bucken Gobbi / Banca: José Angelo Barela / Banca: Sérgio Teixeira Fonseca / Resumo: Durante a locomoção, ocorre um padrão de movimento visual referido como um campo de fluxo óptico. Para explorar como os ajustamentos locomotores são influenciados por este padrão, um paradigma experimental foi desenvolvido para anular o fluxo óptico durante a locomoção e ultrapassagem de obstáculo. O objetivo deste estudo foi investigar a contribuição de amostragens visuais estáticas e dinâmica na locomoção e ultrapassagem de obstáculos de diferentes alturas. Dez indivíduos (23,4 l 1,28 anos de idade, de ambos os gêneros) foram convidados a andar sobre uma passarela e ultrapassar um de dois obstáculos personalizados (obstáculo alto = altura do joelho e obstáculo baixo = altura do tornozelo) posicionado a 5m do ponto de partida. Diodos emissores de luz foram afixados no 5º metatarso e face lateral do calcâneo em ambos os pés e suas trajetórias foram filmados por quatro filmadoras digitais posicionadas bilateralmente à passarela. Dois blocos de tentativas foram apresentados em função das condições de amostragens visuais. No primeiro bloco, 10 tentativas com amostragem visual dinâmica (iluminação normal) foram apresentadas. No segundo bloco de tentativas, duas condições de amostragens visuais estáticas foram apresentadas de forma aleatória por meio de uma luz estroboscópica (2 flashes/segundo e 4 flashes/segundo). A duração de cada flash não ultrapassou 16ms e forneceu apenas imagens estáticas do ambiente. A altura do obstáculo foi randomizada em cada bloco de tentativas. Cinco tentativas para cada condição foram coletadas, totalizando 30 tentativas por participante. Todas as imagens foram digitalizadas e reconstruídas no plano tri-dimensional. As variáveis dependentes na fase de aproximação (comprimentos dos dois passos anteriores à ultrapassagem) e na fase de ultrapassagem... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: During locomotion, a pattern of visual motion referred as an optic flow field occurs. In order to explore how the locomotor adjustments are influenced by this pattern, an experimental paradigm was developed to remove the optic flow during locomotion and obstacle avoidance. The aim of this study was to investigate the contribution of the static and dynamic visual sampling on the locomotion and obstacle avoidance of different heights. Ten individuals (23.4 l 1.28 years of age, both genders) were invited to walk on a pathway and to avoid one out of two personalized obstacles (high obstacle = knee height and low obstacle = ankle height) positioned 5m from the starting position. Light emitting diodes were fixed on the fifth metatarsal and calcaneus lateral face in both feet and their trajectories were recorded by four digital cameras bilaterally positioned related to the pathway. Two trial blocks were presented depending of the visual sampling conditions. In the first block, 10 trials with visual dynamic sampling (normal light) were presented. In the second trial block, two conditions of the static visual sampling frequencies were presented in an randomized way by means of the stroboscopic light (2 flashes/second or 4 flashes/second). Each flash duration was about 16ms and provided only environmental static images. Obstacle height was randomized for each trial block. Five trials for each experimental condition were collected totalizing 30 trials for each participant. All images were digitalized and reconstructed in the 3D plane. Dependent variables in the approach phase (last two step lengths before crossing) and in the crossing phase (horizontal toe distance to the obstacle prior crossing, obstacle toe clearance and mean horizontal velocity for each foot) were selected. The results revealed differences... (Complete abstract, access undermentioned electronic address) / Mestre