Neurobiological and neuroethological findings on insects can be used to design and construct small robots controlling their navigation on the basis of bio-inspired visual strategies and circuits. Animals' visual guidance is partly mediated by motion-sensitive neurons, which are responsible for gauging the optic flow. Although neurons of this kind were discovered in vertebrates' and invertebrates' visual systems more than 50 years ago, the principles and neural mechanisms involved have not yet been completely elucidated. Here, first, I propose to outline some of the findings we made during the last few decades by performing electrophysiological recordings on identified neurons in the housefly's eye while applying optical stimulation to identified photoreceptors. Whereas these findings shed light on the inner processing structure of an elementary motion detector (EMD), recent studies in which the latest genetic and neuroanatomical methods were applied to the fruitfly's visual system have identified some of the neurons in the visual chain which are possibly involved in the neural circuitry underlying a given EMD. Then, I will describe some of the proof-of-concept robots that we have developed on the basis of our biological findings. The 100-g robot OCTAVE, for example, is able to avoid the ground, react to wind, and land autonomously on a flat terrain without ever having to measure any state variables such as distances or speeds. The 100-g robots OSCAR 1 and OSCAR 2 inspired by the microscanner we discovered in the housefly's eye are able to stabilize their body using mainly visual means and track a moving edge with hyperacuity. These robots react to the optic flow, which is sensed by miniature optic flow sensors inspired by the housefly's EMDs. Constructing a “biorobot” gives us a unique opportunity of checking the soundness and robustness of a principle that is initially thought to be understood by bringing it face to face with the real physical world. Bio-inspired robotics not only help neurobiologists and neuroethologists to identify and investigate worthwhile problems in animals' sensory-motor systems, but they also provide engineers with ideas for developing novel devices and machines with promising future applications, in the field of smart autonomous vehicles and microvehicles, for example.

OSCAR II is a twin-engine aerial demonstrator equipped with a monocular visual system, which manages to keep its gaze and its heading steadily fixed on a target (i.e., a dark edge or a bar) in spite of the severe random perturbations applied to its body via a ducted fan. The tethered robot stabilizes its gaze on the basis of two oculomotor reflexes (ORs) inspired by studies on animals: 1) a visual-fixation reflex (VFR) and 2) a vestibulo-ocular reflex (VOR). One of the key features of this robot is that the eye is decoupled mechanically from the body about the vertical (i.e., yaw) axis. To meet the conflicting requirements of high accuracy and fast ocular responses, a miniature (2.4 g) voice-coil motor (VCM) was used, which enables the eye to make a change of orientation with an unusually short rise time (19 ms). The robot, which was equipped with a high-bandwidth (7 Hz) ¿VOR,¿ which is based on an inertial microrate gyro, is capable of accurate visual fixation as long as there is light. The robot is also able to pursue a moving target in the presence of erratic gusts of wind. Here, we present the two interdependent control schemes driving the eye in the robot and the robot in space with no knowledge of the robot's angular position. This ¿steering-by-gazing¿ control strategy, which is implemented on this lightweight (100 g) miniature aerial robot, demonstrates the effectiveness of this biomimetic visual/inertial heading control strategy.

We have developed a new vision based autopilot able to make a micro-air vehicle automatically navigate in steep relief. It uses onboard Optic Flow sensors inspired by the housefly's Elementary Motion Detectors (EMDs) that were previously investigated at our Laboratory with electrophysiological and micro-optical techniques. The paper investigates how the ground avoidance performances of the former OCTAVE robot could be enhanced to cope with steep relief. The idea is to combine frontal and ventral OF sensors and to merge feedback and feedforward loops. In the new robot, called OCTAVE(2), a feedback loop adjusts the lift so as to keep the ventral OF constant, while a feedforward loop based on a forward looking EMD sensor serves to anticipate the steep relief. We test the new autopilot on a 100-gram tethered rotorcraft that circles indoors over an environment composed of contrasting features randomly arranged on the floor. We show that OCTAVE(2) succeeds in following a relatively steep relief (maximum slope 17deg) while navigating close to the ground (groundheight in the order of lm). This risky task is performed thanks to a minimalist electronic visual system: the OF sensor suite is lightweight (4.3-grams including the lenses), and is therefore mounted onboard.

Autonomous guidance of micro-air vehicles (MAVs) in unknown environments is a challenging task because these artificial creatures have small aeromechanical time constants, which make them prone to be disturbed by gusts of wind. Flying insects are subject to quite similar kinds of disturbances, yet they navigate swiftly and deftly. Flying insects display high- performance visuo-motor control systems that have stood the test of time. They can therefore teach us how vision can be used for immediate and vital actions. We built a 50-gram tethered aerial demonstrator, called OSCAR II, which manages to keep its gaze steadily fixating a target (a dark edge), in spite of nasty thumps that we deliberately gave to its body with a custom-made "slapping machine". The robot's agile yaw reactions are based on: (1) a mechanical decoupling of the eye from the body, (2) an active coupling of the robot's heading with its gaze, (3) a Visual Fixation Reflex (VFR), (4) a Vestibulo-Ocular Reflex (VOR), (5) an accurate and fast actuator (voice coil motor, VCM). The actuator is a 2.4-gram voice coil motor that is able to rotate the eye with a rise time as small as 12 ms, that is, much shorter than the rise time of human oculo-motor saccades. In connection with a micro-rate gyro, this actuator endows the robot with a high performance "vestibulo ocular reflex" that keeps the gaze locked onto the target whatever perturbations in yaw affect the robot's body. Whenever the robot is destabilized (e.g., by a slap applied on one side), the gaze keeps fixating the target, while being the reference to which the robot's heading is servoed. It then takes the robot only 0.6 s to realign its heading with its gaze.

We have developed a visually based autopilot which is able to make a micro air vehicle (MAV) automatically take off, cruise and land, while reacting adequately to wind disturbances. We built a proof-of-concept, tethered rotorcraft that can travel indoors over an environment composed of contrasting features randomly arranged on the floor. Here we show the feasibility of a visuomotor control loop that acts upon the thrust so as to maintain the optic flow (OF) estimated in the downward direction to a reference value. The sensor involved in this OF regulator is an elementary motion detector (EMD). The functional structure of the EMD was inspired by that of the housefly, which was previously investigated at our laboratory by performing electrophysiological recordings while applying optical microstimuli to single photoreceptor cells of the compound eye. The vision based autopilot, which we have called OCTAVE (optic flow control system for aerospace vehicles) solves complex problems such as terrain following, controls risky maneuvers such as take off and landing and responds appropriately to wind disturbances. All these reputedly demanding tasks are performed with one and the same visuomotor control loop. The non-emissive sensor and simple processing system are particularly suitable for use with MAV, since the tolerated avionic payload of these micro-aircraft is only a few grams. OCTAVE autopilot could also contribute to relieve a remote operator from the lowly and difficult task of continuously piloting and guiding an UAV. It could also provide guiding assistance to pilots of manned aircraft.

For any sighted mobile creature, whether it be natural or artificial, stabilizing the visual system is a much more crucial issue than that of preventing a snapshot from being blurred by the unsteadiness of a human photographer. The more immune the eye of an animal or a robot is to various kind of disturbances such as body or head movements of all kinds, the less troublesome it will be for the visual system to carry out its many information processing tasks when walking or flying in unknown environments. The gaze control system that we describe In this work takes a lesson from the vestibulo ocular reflex (VOR) that is known to contribute to the stabilization of the human eyes. The originality and performances of the control system arise from the merging of two sensory modalities: (a) a retinal position signal is yielded by a novel piezo-based visual sensor called OSCAR (optical sensor for the control of autonomous robots); and (b) a VOR reflex is merged with a visual smooth pursuit reflex. Our gaze controller involves a feedforward eye control based on measurements of the angular head speed by a rate gyro. The performances of the gaze controller were tested on-board a miniature (5 grams) oculomotor system, which makes use of the OSCAR visual sensor. The combined visual/inertial sensory-motor controller enables the gaze to be stabilized within a 12-times smaller range than the perturbing head movements, which were applied here at frequencies of up to 3 Hz with an amplitude of 6/spl deg/ peak-to-peak. This is a relatively high standard of performance in terms of rejecting head movement disturbances; in any case, these performances are comparable to those which the human visual system is capable of.

Describes a sensor designed to ensure the non-contact stabilization of a craft and especially to deal with disturbance problems such as those resulting from slow speed angular drift. A low-cost, low-complexity active vision system is described, which is based on the specific eye movements occurring in the compound eye of the fly. In our system, motion is detected and processed by a local motion detector circuit (LMD). First, the rotation of two photosensors at a constant angular speed was simulated, which emphasized motion processing as a useful means of detecting variably contrasted objects, however far ahead they are located. Secondly, we reasoned that if the pair of photosensors turns at a varying angular speed the signal delivered by the LMD will vary depending on the position of the contrast feature located in the sensor's visual field. The validity of this reasoning was then tested by constructing a miniature scanning sensor, the output voltage of which turned out to be a quasi-linear function of the position of the contrast feature present in the visual field and to be largely independent of the distance and the level of contrast. This output can therefore be used to generate the appropriate motor commands for stabilizing a sensory platform subject to yaw, pitch, or roll in relation to environmental features and/or for tracking contrasting objects.

Summary form only given. From the knowledge we gained about the processing of motion in the fly visual system we explored how an agent can use motion vision to control its locomotion. We built a biologically inspired visual system made of opto-electronic motion detecting "neurons". Instead of involving a von Neumann architecture, this system is based on parallel and analog networks, as is the case with nervous systems in general. This visual system now guides a mobile robot without collision in unforeseen environments. Computation taking place onboard our robot essentially relies on brainlike, parallel, analogue, continuous time, asynchronous networks, with many additional features common to advanced animal visual systems, such as nonuniform retinal sampling, retinotopic projections of sensory maps, saccadic suppression, and corollary discharges that inhibit vision during eye rotations. All circuits are hardwired, yet adaptive behavior is achieved in the sense that the robot immediately copes with novel environments.