Robust Position Estimation of an Autonomous Mobile Robot

摘要: In the past few years, topic of localization has received considerable attention in research community and especially mobile robotics area (Borenstein, 1996). It consists estimating robot’s pose (position, orientation) with respect to its environment from sensor data. Therefore, better sensory data exploitation is required increase autonomy. The simplest way estimate parameters integration odometric which, however, associated unbounded errors, resulting uneven floors, wheel slippage, limited resolution encoders, etc. However, such a technique not reliable due cumulative errors occurring over long run. robot must also be able localize or internal world model by using information obtained external sensors. system localization, use range disparate multiple sensors, automatically extract maximum amount possible about sensed under all operating conditions. Usually, for many problems like obstacle detection, Simultaneous Localization Map Building (SLAM) (Montemerlo et al., 2002), perception relies on fusion several kinds sensors video cameras, radars, dead-reckoning multi-sensor problem popularly described state space equations defining interesting state, evolution observation models. Based this description, estimation can formulated as tracking problem. To deal problem, when uncertainty occurs, probabilistic Bayesian approaches are most used robotics, even if new set-membership one (Gning & Bonnifait, 2005) Belief theory (Ristic Smets, 2004) have proved themselves some applications. SLAM robots build up map within an unknown while at same time keeping track their current position. Several works implementing algorithms been studied extensively last years direction, leading that classified into three well differentiated paradigms depending underlying structure: metric (Sim 2006) (Tardos