Nathan Sprague

Modeling embodied visual behaviors

To make progess in understanding human visuomotor behavior, we will need to understand its basic components at an abstract level. One way to achieve such an understanding would be to create a model of a human that has a sufficient amount of complexity so as to be capable of generating such behaviors. Recent technological advances have been made that allow progress to be made in this direction. Graphics models that simulate extensive human capabilities can be used as platforms from which to develop synthetic models of visuomotor behavior. Currently, such models can capture only a small portion of a full behavioral repertoire, but for the behaviors that they do model, they can describe complete visuomotor subsystems at a useful level of detail. The value in doing so is that the bodys elaborate visuomotor structures greatly simplify the specification of the abstract behaviors that guide them. The net result is that, essentially, one is faced with proposing an embodied “operating system” model for picking the right set of abstract behaviors at each instant. This paper outlines one such model. A centerpiece of the model uses vision to aid the behavior that has the most to gain from taking environmental measurements. Preliminary tests of the model against human performance in realistic VR environments show that main features of the model show up in human behavior.

Clothed people detection in still images

We present a trainable system for locating clothed people in photographic images. People detection is a particularly challenging image understanding problem; as a result of variations in clothing and posture, the appearance of people may vary enormously from image to image. Our approach attempts to construct a maximally person-like assembly of image regions, where candidate regions are provided by color-based segmentation followed by non-purposive grouping. A tree structured probability model is employed to allow efficient searches. This structure represents the pairwise configuration of body parts as a function of relative position, relative size, and adjacency. Face and skin detection is also used to help the search. The problem of occlusion is addressed through a mixture of trees, where the different mixture components represent the possible subsets of visible parts. Different clothing styles are accounted for by separate models. Experimental results are shown to demonstrate the promise of and challenges for the current system.

Basis iteration for reward based dimensionality reduction

We propose a linear dimensionality reduction algorithm that selectively preserves task relevant state data for control problems modeled as Markov decision processes. The algorithm works by alternating value function estimation with basis vector adaptation. The approach is demonstrated on two tasks: a toy task designed to illustrate the key concepts, and a more complex three dimensional navigation task.

Modeling the brain's operating system

To make progess in understanding human brain functionality, we will need to understand its basic functions at an abstract level. One way of accomplishing such an integration is to create a model of a human that has a useful amount of complexity. Essentially, one is faced with proposing an embodied “operating system” model that can be tested against human performance. Recently technological advances have been made that allow progress to be made in this direction. Graphics models that simulate extensive human capabilities can be used as platforms from which to develop synthetic models of visuo-motor behavior. Currently such models can capture only a small portion of a full behavioral repertoire, but for the behaviors that they do model, they can describe complete visuo-motor subsystems at a level of detail that can be tested against human performance in realistic environments. This paper outlines one such model and shows both that it can produce interesting new hypotheses as to the role of vision and also that it can enhance our understanding of visual attention.

Contingent Features for Reinforcement Learning

Applying reinforcement learning algorithms in real-world domains is challenging because relevant state information is often embedded in a stream of high-dimensional sensor data. This paper describes a novel algorithm for learning task-relevant features through interactions with the environment. The key idea is that a feature is likely to be useful to the degree that its dynamics can be controlled by the actions of the agent. We describe an algorithm that can find such features and we demonstrate its effectiveness in an artificial domain.

Teaching Robotics Using ROS (Abstract Only)

The deployment of autonomous and semi-autonomous robots is likely to increase dramatically over the next decade. Recent autonomous vehicle prototypes illustrate both the rapid progress of the underlying technology and the commercial possibilities of robotics. The next few years are likely to see increased interest in robotics among both students and employers. The Robot Operating System (ROS) is an open-source software framework for developing robotics applications. It has become a standard platform with a wide range of supported robots and a vibrant software ecosystem. This workshop will provide a hands-on introduction to ROS. Participants will have the opportunity to write ROS-based Python programs to control a Turtlebot educational robot. We will discuss the benefits and challenges of using ROS in an undergraduate robotics course. The workshop is intended for CS educators with an interest in teaching robotics. Laptops will be provided. No experience with ROS or Python is required.