Egocentric Direction and the Visual Guidance of Robot Locomotion Background, Theory and Implementation

Rushton, Simon K.; Wen, Jia; Allison, Robert S.

doi:10.1007/3-540-36181-2_58

Cited by 14 publications

(12 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recall as well that the modeling strategy is to fix these values and predict behavior in more complex environments with no free parameters. Rushton, Wen, and Allison (2002) proposed but did not test a simpler hypothesis for steering to a goal in which, rather than bringing the heading error to zero, the agent holds the goal at a fixed eccentricity from the body's midline, corresponding to one free parameter. This generates paths in the form of equiangular spirals, with greater curvature (higher turning rate) near the end of the trajectory, as the target is approached.…”

Section: Steering To a Goalmentioning

confidence: 99%

The dynamics of perception and action.

Warren¹

2006

Psychological Review

689

554

View full text Add to dashboard Cite

How might one account for the organization in behavior without attributing it to an internal control structure? The present article develops a theoretical framework called behavioral dynamics that integrates an information-based approach to perception with a dynamical systems approach to action. For a given task, the agent and its environment are treated as a pair of dynamical systems that are coupled mechanically and informationally. Their interactions give rise to the behavioral dynamics, a vector field with attractors that correspond to stable task solutions, repellers that correspond to avoided states, and bifurcations that correspond to behavioral transitions. The framework is used to develop theories of several tasks in which a human agent interacts with the physical environment, including bouncing a ball on a racquet, balancing an object, braking a vehicle, and guiding locomotion. Stable, adaptive behavior emerges from the dynamics of the interaction between a structured environment and an agent with simple control laws, under physical and informational constraints.Keywords: perception and action, perceptual-motor control, dynamical systems, self-organization, locomotionThe organization of behavior has been a central concern of psychology for well over a century. How is it that humans and other animals can generate behavioral patterns that are tightly coordinated with the environment, in the service of achieving a specific goal? This ability to produce stable yet adaptive behavior raises two constituent issues. On the one hand, it implicates the coordination of action, such that the many neuromusculoskeletal components of the body become temporarily organized into an ordered pattern of movement. On the other, it implicates perception, such that information about the world and the body enables appropriate actions to be selected and adapted to environmental conditions. At a basic level, the problem of the organization of behavior is thus synonymous with the problem of perception and action. Moreover, an adequate theory of perceptually controlled action would provide a platform for understanding more "cognitive" behavior such as extended action sequences, anticipatory behavior oriented to remote goals, or predictive behavior that takes account of hidden environmental properties.It seems natural to presume that observed organization in behavior implies ipso facto the existence of a centralized controller-a pattern generator, action plan, or internal model that is responsible for its organization and regulation. Such an assumption has been commonplace in psychology, cognitive science, neuroscience, and robotics. In each domain, organization in behavior has been attributed to prior organization in the structure of the nervous system (the neuroreductionist view), the structure of internal representations (the cognitivist view), or in the contingencies presented by the environment (the behaviorist view). This is unsatisfying because it merely displaces the original problem of behavioral organization to a preexist...

show abstract

Section: Steering To a Goalmentioning

confidence: 99%

The dynamics of perception and action.

Warren¹

2006

Psychological Review

689

554

View full text Add to dashboard Cite

show abstract

“…It can be seen from the trajectories in Figure 1 that human participants actually steer more directly to the target than would be predicted if they simply nulled target drift. Rushton et al (2002) proposed that more direct paths could be generated by overcompensating for target drift (e.g. steering at 200% target drift).…”

mentioning

confidence: 99%

Active gaze, visual look-ahead, and locomotor control.

Wilkie¹,

Wann²,

Allison³

2008

Journal of Experimental Psychology: Human Perception and Perfor

Self Cite

162

View full text Add to dashboard Cite

We examined observers steering through a series of obstacles to determine the role of active gaze in shaping locomotor trajectories. Participants sat on a bicycle trainer integrated with a large field of view simulator and steered through a series of slalom gates. Steering behavior was determined by examining the passing distance through gates and the smoothness of trajectory. Gaze monitoring revealed which slalom targets were fixated and for how long. Participants tended to track the most immediate gate until it was ~1.5 seconds away at which point gaze switched to the next slalom gate. To probe this gaze pattern we then introduced a number of experimental conditions that placed spatial or temporal constraints upon where participants could look and when. These manipulations resulted in systematic steering errors when observers were forced to use unnatural looking patterns, but errors reduced when peripheral monitoring of obstacles was allowed. We propose a steering model based upon active gaze sampling informed by our experimental conditions and consistent with our observations in free gaze experiments and with recommendations from real world high-speed steering.Keywords: Locomotion, Steering, Gaze, Eye Movements, Active Vision Please direct correspondence to: Dr. Richard M.Wilkie, Institute of Psychological Sciences, The University of Leeds, Leeds, LS2 9JT, England. Email: r.m.wilkie@leeds.ac.uk Web: www.personal.leeds.ac.uk/~pscrmw/ Thanks to the Centre for Vision Research, University of York, Toronto for their support during the collaborative research visit that contributed to the ideas presented here, and also thanks to Mike Harris and two anonymous reviewers for constructive input into earlier versions of this manuscript. Research supported by the UK EPSRC GR/S86358 and EP/D055342/1. This article may not exactly replicate the final version published in the Journal of Experimental Psychology: Human Perception and Performance. It is not the copy of record. WILKIE, WANN & ALLISON -ACTIVE GAZE, VISUAL LOOK-AHEAD AND LOCOMOTOR CONTROL 2Active visual exploration is a crucial part of interacting with our environment and accurately perceiving the world around us (Wexler & van Boxtel, 2005). Effective locomotion is initiated through a series of online control movements, which are generated by a fast and efficient perception-action loop. This response system involves movement distributed across the body invoking eye, head and whole body motion, which in turn affects the online information that is available to guide steering (Wilkie & Wann, 2002;. Land and Lee (1994), and Land and Tatler (2001) highlighted the use of head and gaze orienting in directing steering during car driving, however, such behaviors could provide two sources of information stemming directly from the orienting action: gaze angle, (the direction of your eyes and head relative to your body"s midline, which is usually coincident with your direction of travel), and also gaze rotation. Gaze angle (visual direction) gives some indication of the magnit...

show abstract

“…By comparison, pure simulations and ideal-observer analyses are totally determined by the input parameters. In this respect designing a robot that will move autonomously through the world focuses the issue of what informational variables are critical to the visual control of locomotion and serves to outline what is necessary and what is sufficient [Rushton et al 2002].…”

Section: Why Build a Locomotor Robot?mentioning

confidence: 99%

“…Maintaining a constant target angle is a strategy used by raptors when approaching prey [Tucker 2000] who break out of the spiral near the end of the flight to take a rapid straight path to their prey. A constant α model has also been demonstrated to be effective for controlling a locomotor robot [Rushton et al 2002]. In humans, when steering to a target, the rate of steering depends upon the quality of information that is made available (as discussed later), however humans tend to follow a course that turns more quickly than merely keeping α constant or Equation 1 (Figure 2, Left Panel).…”

Section: Active Gaze Models Of Steeringmentioning

confidence: 99%

“…Here we test this model using the robotic platform. Having the camera at a fixed height above a flat ground plane means that when a steering target is fixated, the angle of declination becomes a useful estimate of target immediacy [Rushton et al 2002]. The vertical angle of a point on the camera image is directly related to the distance of that point in the world.…”

Section: Experiments 3: Steering Via Multiple Targetsmentioning

confidence: 99%

See 1 more Smart Citation

Modeling locomotor control

Wilkie

Wann

Allison

2011

ACM Trans. Appl. Percept.

Self Cite

View full text Add to dashboard Cite

In 1958 JJ Gibson put forward proposals on the visual control of locomotion. Research in the last 50 years has served to clarify the sources of visual and non-visual information which contribute to successful steering, but has yet to determine how this information is optimally combined under conditions of uncertainty. Here we test the conditions under which a locomotor robot with a mobile camera can steer effectively using simple visual and extra-retinal parameters to examine how such models cope with the noisy real-world visual and motor estimates that are available to humans. This applied modelling gives us an insight into both the advantages and limitations of using active gaze to sample information when steering.

show abstract

Egocentric Direction and the Visual Guidance of Robot Locomotion Background, Theory and Implementation

Cited by 14 publications

References 26 publications

The dynamics of perception and action.

The dynamics of perception and action.

Active gaze, visual look-ahead, and locomotor control.

Modeling locomotor control

Contact Info

Product

Resources

About