Orientador: Renata Cardoso Magagnin / Banca: João Roberto Gomes de Faria / Banca: Maria Leonor Alves Maia / Resumo: A busca das cidades contemporâneas em implantar o conceito de mobilidade urbana sustentável apoia-se, dentre outros aspectos, na implementação de infraestrutura que possibilite o deslocamento a pé, ou caminhada. No Brasil, muitos municípios que querem implantar este conceito têm discutido a qualidade da infraestrutura ofertada aos pedestres (também denominada de caminhabilidade), em especial, aquelas pessoas que possuem alguma deficiência ou mobilidade reduzida. Diante desse contexto, o objetivo dessa pesquisa é propor uma ferramenta para análise da caminhabilidade que possibilite avaliar o entorno de polos geradores de viagem, englobando a macro e micro escala de análise urbana, e a percepção da população e de pesquisadores e gestores. O instrumento foi aplicado no município de Uberlândia (MG), no entorno do Terminal Central Paulo Ferolla da Silva. A metodologia foi composta por: i) definição do Índice de Caminhabilidade de Macro e Micro Escala (ICMME) e ii) aplicação do instrumento. Os resultados apontam para a efetividade do instrumento em identificar os principais problemas no local avaliado e em gerar diretrizes de melhoria da caminhabilidade, bem como na definição de critérios considerados mais importantes na escolha em andar a pé. Para futuros trabalhos, recomenda-se a aplicação do índice em outros recortes espaciais para avaliação da caminhabilidade nas cidades, por pesquisadores ou pela administração pública. / Abstract: The search of contemporary cities to implement the concept of sustainable urban mobility is supported, among other aspects, by the implementation of infrastructure that enables walking. In Brazil, many municipalities that want to implement this concept have discussed the quality of infrastructure offered to pedestrians (also known as walkability), especially those with disabilities or reduced mobility. In this context, the objective of this research is to propose a tool for walkability analysis that evaluates the surroundings of travel generating poles, encompassing the macro and micro scale of urban analysis, and the perception of population and of researchers and managers. The instrument was applied in the city of Uberlândia (MG), around the Terminal Paulo Ferolla da Silva Central. The methodology consisted of: i) definition of the Macro and Micro Scale Walkability Index (in portuguese ICMME) and ii) application of the instrument. The results point to the effectiveness of the instrument in identifying the main problems in the evaluated place and in generating guidelines for improving walkability, as well as in the definition of criteria considered more important in choosing to walk. For future researchers, it is recommended to apply the index in other space clippings for evaluating walkability in the cities, by researchers or by the public administration. / Mestre

The purpose of this dissertation is to address the problem of predicting pedestrian movement and behavior in and among crowds. Specifically, we will focus on an agent based approach where pedestrians are treated individually and parameters for an energy model are trained by real world video data. These learned pedestrian models are useful in applications such as tracking, simulation, and artificial intelligence. The applications of this method are explored and experimental results show that our trained pedestrian motion model is beneficial for predicting unseen or lost tracks as well as guiding appearance based tracking algorithms. The method we have developed for training such a pedestrian model operates by optimizing a set of weights governing an aggregate energy function in order to minimize a loss function computed between a model's prediction and annotated ground-truth pedestrian tracks. The formulation of the underlying energy function is such that using tight convex upper bounds, we are able to efficiently approximate the derivative of the loss function with respect to the parameters of the model. Once this is accomplished, the model parameters are updated using straightforward gradient descent techniques in order to achieve an optimal solution. This formulation also lends itself towards the development of a multiple behavior model. The multiple pedestrian behavior styles, informally referred to as "stereotypes", are common in real data. In our model we show that it is possible, due to the unique ability to compute the derivative of the loss function, to build a new model which utilizes a soft-minimization of single behavior models. This allows unsupervised training of multiple different behavior models in parallel. This novel extension makes our method unique among other methods in the attempt to accurately describe human pedestrian behavior for the myriad of applications that exist. The ability to describe multiple behaviors shows significant improvements in the task of pedestrian motion prediction.