No abstract

Rats traversing a straight alley once a day for delivery of a single i.v. injection of cocaine develop over trials an ambivalence about entering the goal box. This ambivalence is characterized by the increasing occurrence of "retreat behaviors" where animals leave the start box and run quickly to the goal box, but then stop at the entry point and "retreat" back toward the start box. This unique pattern of retreat behavior has been shown to reflect a form of "approach-avoidance conflict" that stems from the animals' concurrent positive (cocaine reward) and negative (cocaine-induced anxiety) associations with the goal box. Cocaine blocks reuptake of the serotonergic (5-HT) transporter and serotonin has been implicated in the modulation of anxiety. It was therefore of interest to determine whether inactivation of the serotonergic cell bodies residing in the dorsal raphé nucleus (DRN) and projecting to brain areas critical for the modulation of anxiety, would alter the anxiogenic state exhibited by rats running an alley for single daily i.v. injections of 1.0 mg/kg cocaine. Reversible inactivation of the DRN was accomplished by intracranial application of a mixed solution of the GABA agonists baclofen and muscimol. While DRN inactivation had no impact on the subjects' motivation to initiate responding (i.e., latencies to leave the start box were unaffected) it reliably reduced the frequency of approach-avoidance retreat behaviors (conflict behavior). These data suggest that inactivation of the dorsal raphé reduces the conflict/ anxiety otherwise present in experienced cocaine-seeking animals.

Learning from Demonstration (LfD) enables novice users to teach robots new skills. However, many LfD methods do not facilitate skill maintenance and adaptation. Changes in task requirements or in the environment often reveal the lack of resiliency and adaptability in the skill model. To overcome these limitations, we introduce ARC-LfD: an Augmented Reality (AR) interface for constrained Learning from Demonstration that allows users to maintain, update, and adapt learned skills. This is accomplished through insitu visualizations of learned skills and constraint-based editing of existing skills without requiring further demonstration. We describe the existing algorithmic basis for this system as well as our Augmented Reality interface and the novel capabilities it provides. Finally, we provide three case studies that demonstrate how ARC-LfD enables users to adapt to changes in the environment or task which require a skill to be altered after initial teaching has taken place.

This thesis summary presents research focused on incorporating high-level abstract behavioral requirements, called 'conceptual constraints', into the modeling processes of robot Learning from Demonstration (LfD) techniques. This idea is realized via an LfD algorithm called Concept Constrained Learning from Demonstration. This algorithm encodes motion planning constraints as temporally associated logical formulae of Boolean operators that enforce highlevel constraints over portions of the robot's motion plan during learned skill execution. This results in more easily trained, more robust, and safer learned skills. Current work focuses on automating constraint discovery, introducing conceptual constraints into human-aware motion planning algorithms, and expanding upon trajectory alignment techniques for LfD. Future work will focus on how concept constrained algorithms and models are best incorporated into effective interfaces for end-users. CCS CONCEPTS• Computing methodologies → Learning from demonstrations; Robotic planning; • Computer systems organization → Robotic autonomy; External interfaces for robotics.