Andrew M. Saxe

Acquisition of decision making criteria: reward rate ultimately beats accuracy

Speed–accuracy trade-offs strongly influence the rate of reward that can be earned in many decision-making tasks. Previous reports suggest that human participants often adopt suboptimal speed–accuracy trade-offs in single session, two-alternative forced-choice tasks. We investigated whether humans acquired optimal speed–accuracy trade-offs when extensively trained with multiple signal qualities. When performance was characterized in terms of decision time and accuracy, our participants eventually performed nearly optimally in the case of higher signal qualities. Rather than adopting decision criteria that were individually optimal for each signal quality, participants adopted a single threshold that was nearly optimal for most signal qualities. However, setting a single threshold for different coherence conditions resulted in only negligible decrements in the maximum possible reward rate. Finally, we tested two hypotheses regarding the possible sources of suboptimal performance: (1) favoring accuracy over reward rate and (2) misestimating the reward rate due to timing uncertainty. Our findings provide support for both hypotheses, but also for the hypothesis that participants can learn to approach optimality. We find specifically that an accuracy bias dominates early performance, but diminishes greatly with practice. The residual discrepancy between optimal and observed performance can be explained by an adaptive response to uncertainty in time estimation.

Kratos: Princeton University's entry in the 2008 Intelligent Ground Vehicle Competition

In this paper, we present Argos, an autonomous ground robot built for the 2009 Intelligent Ground Vehicle Competition (IGVC). Discussed are the significant improvements over its predecessor from the 2008 IGVC, Kratos. We continue to use stereo vision techniques to generate a cost map of the environment around the robot. Lane detection is improved through the use of color filters that are robust to changing lighting conditions. The addition of a single-axis gyroscope to the sensor suite allows accurate measurement of the robots yaw rate and compensates for wheel slip, vastly improving state estimation. The combination of the D* Lite algorithm, which avoids unnecessary re-planning, and the Field D* algorithm, which allows us to plan much smoother paths, results in an algorithm that produces higher quality paths in the same amount of time as methods utilizing A*. The successful implementation of a crosstrack error navigation law allows the robot to follow planned paths without cutting corners, reducing the chance of collision with obstacles. A redesigned chassis with a smaller footprint and a bi-level design, combined with a more powerful drivetrain, makes Argos much more agile and maneuverable compared to its predecessor. At the 2009 IGVC, Argos placed first in the Navigation Challenge.

Prospect Eleven: Princeton University’s Entry in the 2005 DARPA Grand Challenge

This paper describes Princeton University’s approach to the 2005 DARPA Grand Challenge, an off-road race for fully autonomous ground vehicles. The system, Prospect Eleven, takes a simple approach to address the problems posed by the Grand Challenge including obstacle detection, path planning, and extended operation in harsh environments. Obstacles are detected using stereo vision, and tracked in the time domain to improve accuracy in localization and reduce false positives. The navigation system processes a geometric representation of the world to identify passable regions in the terrain ahead, and the vehicle is controlled to drive through these regions. Performance of the system is evaluated both during the Grand Challenge and in subsequent desert testing. The vehicle completed 9.3 miles of the course on race day, and extensive portions of the 2004 and 2005 Grand Challenge courses in later tests.