Robot navigation in dense crowds : statistical models and experimental studies of human robot cooperation

This thesis explores the problem of mobile robot navigation in dense human crowds. We begin by considering a fundamental impediment to classical motion planning algorithms called the freezing robot problem: once the environment surpasses a certain level of complexity, the planner decides that all forward paths are unsafe, and the robot freezes in place (or performs unnecessary maneuvers) to avoid collisions. Since a feasible path typically exists, this behavior is suboptimal. Existing approaches have focused on reducing predictive uncertainty by employing higher fidelity individual dynamics models or heuristically limiting the individual predictive covariance to prevent overcautious navigation. We demonstrate that both the individual prediction and the individual predictive uncertainty have little to do with this undesirable navigation behavior. Additionally, we provide evidence that dynamic agents are able to navigate in dense crowds by engaging in joint collision avoidance, cooperatively making room to create feasible trajectories. We accordingly develop interacting Gaussian processes, a prediction density that captures cooperative collision avoidance, and a "multiple goal" extension that models the goal driven nature of human decision making. Navigation naturally emerges as a statistic of this distribution.

Most importantly, we empirically validate our models in the Chandler dining hall at Caltech during peak hours, and in the process, carry out the first extensive quantitative study of robot navigation in dense human crowds (collecting data on 488 runs). The multiple goal interacting Gaussian processes algorithm performs comparably with human teleoperators in crowd densities nearing 1 person/m², while a state of the art noncooperative planner exhibits unsafe behavior more than 3 times as often as the multiple goal extension, and twice as often as the basic interacting Gaussian process approach. Furthermore, a reactive planner based on the widely used dynamic window approach proves insufficient for crowd densities above 0.55 people/m². We also show that our noncooperative planner or our reactive planner capture the salient characteristics of nearly any dynamic navigation algorithm. For inclusive validation purposes, we show that either our non-interacting planner or our reactive planner captures the salient characteristics of nearly any existing dynamic navigation algorithm. Based on these experimental results and theoretical observations, we conclude that a cooperation model is critical for safe and efficient robot navigation in dense human crowds.

Finally, we produce a large database of ground truth pedestrian crowd data. We make this ground truth database publicly available for further scientific study of crowd prediction models, learning from demonstration algorithms, and human robot interaction models in general.

Efficient methods for empirical tests of behavioral economics theories in laboratory and field experiments

In the quest for a descriptive theory of decision-making, the rational actor model in economics imposes rather unrealistic expectations and abilities on human decision makers. The further we move from idealized scenarios, such as perfectly competitive markets, and ambitiously extend the reach of the theory to describe everyday decision making situations, the less sense these assumptions make. Behavioural economics has instead proposed models based on assumptions that are more psychologically realistic, with the aim of gaining more precision and descriptive power. Increased psychological realism, however, comes at the cost of a greater number of parameters and model complexity. Now there are a plethora of models, based on different assumptions, applicable in differing contextual settings, and selecting the right model to use tends to be an ad-hoc process. In this thesis, we develop optimal experimental design methods and evaluate different behavioral theories against evidence from lab and field experiments.

We look at evidence from controlled laboratory experiments. Subjects are presented with choices between monetary gambles or lotteries. Different decision-making theories evaluate the choices differently and would make distinct predictions about the subjects' choices. Theories whose predictions are inconsistent with the actual choices can be systematically eliminated. Behavioural theories can have multiple parameters requiring complex experimental designs with a very large number of possible choice tests. This imposes computational and economic constraints on using classical experimental design methods. We develop a methodology of adaptive tests: Bayesian Rapid Optimal Adaptive Designs (BROAD) that sequentially chooses the "most informative" test at each stage, and based on the response updates its posterior beliefs over the theories, which informs the next most informative test to run. BROAD utilizes the Equivalent Class Edge Cutting (EC²) criteria to select tests. We prove that the EC² criteria is adaptively submodular, which allows us to prove theoretical guarantees against the Bayes-optimal testing sequence even in the presence of noisy responses. In simulated ground-truth experiments, we find that the EC² criteria recovers the true hypotheses with significantly fewer tests than more widely used criteria such as Information Gain and Generalized Binary Search. We show, theoretically as well as experimentally, that surprisingly these popular criteria can perform poorly in the presence of noise, or subject errors. Furthermore, we use the adaptive submodular property of EC² to implement an accelerated greedy version of BROAD which leads to orders of magnitude speedup over other methods.

We use BROAD to perform two experiments. First, we compare the main classes of theories for decision-making under risk, namely: expected value, prospect theory, constant relative risk aversion (CRRA) and moments models. Subjects are given an initial endowment, and sequentially presented choices between two lotteries, with the possibility of losses. The lotteries are selected using BROAD, and 57 subjects from Caltech and UCLA are incentivized by randomly realizing one of the lotteries chosen. Aggregate posterior probabilities over the theories show limited evidence in favour of CRRA and moments' models. Classifying the subjects into types showed that most subjects are described by prospect theory, followed by expected value. Adaptive experimental design raises the possibility that subjects could engage in strategic manipulation, i.e. subjects could mask their true preferences and choose differently in order to obtain more favourable tests in later rounds thereby increasing their payoffs. We pay close attention to this problem; strategic manipulation is ruled out since it is infeasible in practice, and also since we do not find any signatures of it in our data.

In the second experiment, we compare the main theories of time preference: exponential discounting, hyperbolic discounting, "present bias" models: quasi-hyperbolic (α, β) discounting and fixed cost discounting, and generalized-hyperbolic discounting. 40 subjects from UCLA were given choices between 2 options: a smaller but more immediate payoff versus a larger but later payoff. We found very limited evidence for present bias models and hyperbolic discounting, and most subjects were classified as generalized hyperbolic discounting types, followed by exponential discounting.

In these models the passage of time is linear. We instead consider a psychological model where the perception of time is subjective. We prove that when the biological (subjective) time is positively dependent, it gives rise to hyperbolic discounting and temporal choice inconsistency.

We also test the predictions of behavioral theories in the "wild". We pay attention to prospect theory, which emerged as the dominant theory in our lab experiments of risky choice. Loss aversion and reference dependence predicts that consumers will behave in a uniquely distinct way than the standard rational model predicts. Specifically, loss aversion predicts that when an item is being offered at a discount, the demand for it will be greater than that explained by its price elasticity. Even more importantly, when the item is no longer discounted, demand for its close substitute would increase excessively. We tested this prediction using a discrete choice model with loss-averse utility function on data from a large eCommerce retailer. Not only did we identify loss aversion, but we also found that the effect decreased with consumers' experience. We outline the policy implications that consumer loss aversion entails, and strategies for competitive pricing.

In future work, BROAD can be widely applicable for testing different behavioural models, e.g. in social preference and game theory, and in different contextual settings. Additional measurements beyond choice data, including biological measurements such as skin conductance, can be used to more rapidly eliminate hypothesis and speed up model comparison. Discrete choice models also provide a framework for testing behavioural models with field data, and encourage combined lab-field experiments.

Quantifying Earthquake Collapse Risk of Tall Steel Moment Frame Buildings Using Rupture-to-Rafters Simulations

There is a sparse number of credible source models available from large-magnitude past earthquakes. A stochastic source model generation algorithm thus becomes necessary for robust risk quantification using scenario earthquakes. We present an algorithm that combines the physics of fault ruptures as imaged in laboratory earthquakes with stress estimates on the fault constrained by field observations to generate stochastic source models for large-magnitude (Mw 6.0-8.0) strike-slip earthquakes. The algorithm is validated through a statistical comparison of synthetic ground motion histories from a stochastically generated source model for a magnitude 7.90 earthquake and a kinematic finite-source inversion of an equivalent magnitude past earthquake on a geometrically similar fault. The synthetic dataset comprises of three-component ground motion waveforms, computed at 636 sites in southern California, for ten hypothetical rupture scenarios (five hypocenters, each with two rupture directions) on the southern San Andreas fault. A similar validation exercise is conducted for a magnitude 6.0 earthquake, the lower magnitude limit for the algorithm. Additionally, ground motions from the Mw7.9 earthquake simulations are compared against predictions by the Campbell-Bozorgnia NGA relation as well as the ShakeOut scenario earthquake. The algorithm is then applied to generate fifty source models for a hypothetical magnitude 7.9 earthquake originating at Parkfield, with rupture propagating from north to south (towards Wrightwood), similar to the 1857 Fort Tejon earthquake. Using the spectral element method, three-component ground motion waveforms are computed in the Los Angeles basin for each scenario earthquake and the sensitivity of ground shaking intensity to seismic source parameters (such as the percentage of asperity area relative to the fault area, rupture speed, and risetime) is studied.

Under plausible San Andreas fault earthquakes in the next 30 years, modeled using the stochastic source algorithm, the performance of two 18-story steel moment frame buildings (UBC 1982 and 1997 designs) in southern California is quantified. The approach integrates rupture-to-rafters simulations into the PEER performance based earthquake engineering (PBEE) framework. Using stochastic sources and computational seismic wave propagation, three-component ground motion histories at 636 sites in southern California are generated for sixty scenario earthquakes on the San Andreas fault. The ruptures, with moment magnitudes in the range of 6.0-8.0, are assumed to occur at five locations on the southern section of the fault. Two unilateral rupture propagation directions are considered. The 30-year probabilities of all plausible ruptures in this magnitude range and in that section of the fault, as forecast by the United States Geological Survey, are distributed among these 60 earthquakes based on proximity and moment release. The response of the two 18-story buildings hypothetically located at each of the 636 sites under 3-component shaking from all 60 events is computed using 3-D nonlinear time-history analysis. Using these results, the probability of the structural response exceeding Immediate Occupancy (IO), Life-Safety (LS), and Collapse Prevention (CP) performance levels under San Andreas fault earthquakes over the next thirty years is evaluated.

Furthermore, the conditional and marginal probability distributions of peak ground velocity (PGV) and displacement (PGD) in Los Angeles and surrounding basins due to earthquakes occurring primarily on the mid-section of southern San Andreas fault are determined using Bayesian model class identification. Simulated ground motions at sites within 55-75km from the source from a suite of 60 earthquakes (Mw 6.0 − 8.0) primarily rupturing mid-section of San Andreas fault are considered for PGV and PGD data.

Real-Time Bayesian Analysis of Ground Motion Envelopes for Earthquake Early Warning

Current earthquake early warning systems usually make magnitude and location predictions and send out a warning to the users based on those predictions. We describe an algorithm that assesses the validity of the predictions in real-time. Our algorithm monitors the envelopes of horizontal and vertical acceleration, velocity, and displacement. We compare the observed envelopes with the ones predicted by Cua & Heaton's envelope ground motion prediction equations (Cua 2005). We define a "test function" as the logarithm of the ratio between observed and predicted envelopes at every second in real-time. Once the envelopes deviate beyond an acceptable threshold, we declare a misfit. Kurtosis and skewness of a time evolving test function are used to rapidly identify a misfit. Real-time kurtosis and skewness calculations are also inputs to both probabilistic (Logistic Regression and Bayesian Logistic Regression) and nonprobabilistic (Least Squares and Linear Discriminant Analysis) models that ultimately decide if there is an unacceptable level of misfit. This algorithm is designed to work at a wide range of amplitude scales. When tested with synthetic and actual seismic signals from past events, it works for both small and large events.