Search

In this work, we present a data-driven simulation and training engine capable of learning end-to-end autonomous vehicle control policies using only sparse rewards. By leveraging real, human-collected trajectories through an environment, we render novel training data that allows virtual agents to drive along a continuum of new local trajectories consistent with the road appearance and semantics, each with a different view of the scene. We demonstrate the ability of policies learned within our simulator to generalize to and navigate in previously unseen real-world roads, without access to any human control labels during training. Our results validate the learned policy onboard a full-scale autonomous vehicle, including in previously un-encountered scenarios, such as new roads and novel, complex, near-crash situations. Our methods are scalable, leverage reinforcement learning, and apply broadly to situations requiring effective perception and robust operation in the physical world.

Evidence from production studies demonstrate that /s/-retraction, a sound change in American English where /s/ approaches [ʃ], is progressing in apparent time and highly context-dependent, occurring overwhelmingly in /str/ clusters. This study reports findings from a phoneme categorization task, suggesting that listeners perceive sibilants in /str/ clusters less categorically than in /spr/ and /skr/ clusters and are becoming less categorical in apparent time for all clusters. Indicators of masculine toughness, including the model talker's voice and face and the listener's endorsement of masculine stereotypes, additionally contribute to a less categorical perception of sibilants in these environments.

Variation, both between and within speakers, is ubiquitous in language. Examining and understanding this variation is crucial not only to questions of socio-linguistics and sound change but also to the study of prosody and phonology more broadly. The present study contributes to the literature on inter and intra-speaker variation in speech production, examining the phonetic realization of prevocalic /s/ in American English using recordings from a longitudinal phonetic corpus of oral arguments before the Supreme Court of the United States. Specifically, this study examines the role of non-segmental factors conditioning the observed variation, focusing on prosodic prominence, segment duration, phonological contrast and lexical frequency. Significant effects for segment duration, speakers’ average /sh/ centroid frequency and word position were observed, with higher centroid frequency values observed in instances of /s/ with a longer relative duration, in word-initial positions, or for speakers with a higher mean /sh/ centroid frequency. There was also significant individual variation in the effects of duration, prosodic prominence and phonological contrast. These results provide further evidence for place of articulation contrast strengthening in prominent positions with novel evidence for place of articulation contrast strengthening in sibilants.

Following CNS injury, myelin-associated inhibitors represent major obstacles to axonal regeneration and functional recovery. The following study suggests that the proteolytic enzyme tissue plasminogen activator (tPA) plays a major function in 'conditioning-injury induced' axon regeneration. In this paradigm, prior peripheral nerve injury leads to an enhanced ability of sensory neurons to regenerate their central axons in the presence of the CNS inhibitory microenvironment. tPA is widely expressed by CNS and PNS neurons and plays major roles in synaptic reorganization and plasticity. This study shows that cultured neurons from mice deficient in tPA, in contrast to wild-type mice, fail to undergo conditioning-injury induced axonal regeneration in the presence of purified myelin membranes. Interestingly, neurons from mice deficient in plasminogen, the best known substrate for tPA, showed active axon regeneration. These results suggest a novel plasminogen-independent role for tPA in promoting axonal regeneration on CNS myelin.