next up previous
Next: Filtering Techniques for Dynamic Up: Metric Markov Localization for Previous: The Action Model

The Perception Model for Proximity Sensors

 

As mentioned above, the likelihood tex2html_wrap_inline2919 that a sensor reading s is measured at position l has to be computed for all positions l in each update of the Markov localization algorithm (see Table 1). Therefore, it is crucial for on-line position estimation that this quantity can be computed very efficiently. [Moravec1988] proposed a method to compute a generally non-Gaussian probability density function tex2html_wrap_inline2919 over a discrete set of possible distances measured by an ultrasound sensor at location l. In a first implementation of our approach [Burgard et al. 1996] we used a similar method, which unfortunately turned out to be computationally too expensive for localization in real-time.

To overcome this disadvantage, we developed a sensor-model which allows to compute tex2html_wrap_inline2919 solely based on the distance tex2html_wrap_inline2985 to the closest obstacle in the map along the direction of the sensor. This distance can be computed by ray-tracing in occupancy grid maps or CAD-models of the environment. In particular, we consider a discretization tex2html_wrap_inline2987 of possible distances measured by a proximity sensor. In our discretization, the size of the ranges tex2html_wrap_inline2989 is the same for all i, and tex2html_wrap_inline2993 corresponds to the maximal range of the proximity sensorgif. Let tex2html_wrap_inline2999 denote the probability of measuring a distance tex2html_wrap_inline3001 if the robot is at location l. In order to derive this probability we first consider the following two cases (see also [Hennig1997, Fox1998]):

  1. Known obstacles: If the sensor detects an obstacle the resulting distribution is modeled by a Gaussian distribution with mean at the distance to this obstacle. Let tex2html_wrap_inline3005 denote the probability of measuring distance d if the robot is at location l, assuming that the sensor beam is reflected by the closest obstacle in the map (along the sensor beam). We denote the distance to this specific obstacle by tex2html_wrap_inline2985 . The probability tex2html_wrap_inline3005 is then given by a Gaussian distribution with mean at tex2html_wrap_inline2985 :

      eqnarray648

    The standard deviation tex2html_wrap_inline3017 of this distribution models the uncertainty of the measured distance, based on

    Figure 4(a) gives examples of such Gaussian distributions for ultrasound sensors and laser range-finders. Here the distance tex2html_wrap_inline2985 to the closest obstacle is 230cm. Observe here that the laser sensor has a higher accuracy than the ultrasound sensor, as indicated by the smaller variance.

      figure1118

  2. Unknown obstacles: In Markov localization, the world model generally is assumed to be static and complete. However, mobile robot environments are often populated and therefore contain objects that are not included in the map. Consequently, there is a non-zero probability that the sensor is reflected by an obstacle not represented in the world model. Assuming that these objects are equally distributed in the environment, the probability tex2html_wrap_inline3029 of detecting an unknown obstacle at distance tex2html_wrap_inline3001 is independent of the location of the robot and can be modeled by a geometric distribution. This distribution results from the following observation. A distance tex2html_wrap_inline3001 is measured if the sensor is not reflected by an obstacle at a shorter distance tex2html_wrap_inline3035 and is reflected at distance tex2html_wrap_inline3001 . The resulting probability is

      eqnarray673

    In this equation the constant tex2html_wrap_inline3039 is the probability that the sensor is reflected by an unknown obstacle at any range given by the discretization.

    A typical distribution for sonar and laser measurements is depicted in Figure 4(b). In this example, the relatively large probability of measuring 500cm is due to the fact that the maximum range of the proximity sensors is set to 500cm. Thus, this distance represents the probability of measuring at least 500cm.

Obviously, only one of these two cases can occur at a certain point in time, i.e., the sensor beam is either reflected by a known or an unknown object. Thus, tex2html_wrap_inline2999 is a a mixture of the two distributions tex2html_wrap_inline3043 and tex2html_wrap_inline3045 . To determine the combined probability tex2html_wrap_inline2999 of measuring a distance tex2html_wrap_inline3001 if the robot is at location l we consider the following two situations: A distance tex2html_wrap_inline3001 is measured, if {

  1. the sensor beam is
    1. not reflected by an unknown obstacle before reaching distance tex2html_wrap_inline3001

        eqnarray695

    2. and reflected by the known obstacle at distance tex2html_wrap_inline3001

        eqnarray700

  2. OR the beam is
    1. reflected neither by an unknown obstacle nor by the known obstacle before reaching distance tex2html_wrap_inline3001

      eqnarray708

    2. and reflected by an unknown obstacle at distance tex2html_wrap_inline3001

        eqnarray712

The parameter tex2html_wrap_inline3063 in Equation (25) denotes the probability that the sensor detects the closest obstacle in the map. These considerations for the combined probability are summarized in Equation (28). By double negation and insertion of the Equations (24) to (27), we finally get Equation (31).

     eqnarray722

To obtain the probability of measuring tex2html_wrap_inline2993 , the maximal range of the sensor, we exploit the following equivalence: The probability of measuring a distance larger than or equal to the maximal sensor range is equivalent to the probability of not measuring a distance shorter than tex2html_wrap_inline2993 . In our incremental scheme, this probability can easily be determined:

  eqnarray731

To summarize, the probability of sensor measurements is computed incrementally for the different distances starting at distance tex2html_wrap_inline3069 cm. For each distance we consider the probability that the sensor beam reaches the corresponding distance and is reflected either by the closest obstacle in the map (along the sensor beam), or by an unknown obstacle.

  figure1152

In order to adjust the parameters tex2html_wrap_inline3017 , tex2html_wrap_inline3039 and tex2html_wrap_inline3063 of our perception model we collected eleven million data pairs consisting of the expected distance tex2html_wrap_inline2985 and the measured distance tex2html_wrap_inline3001 during the typical operation of the robot. From these data we were able to estimate the probability of measuring a certain distance tex2html_wrap_inline3001 if the distance tex2html_wrap_inline2985 to the closest obstacle in the map along the sensing direction is given. The dotted line in Figure 5(a) depicts this probability for sonar measurements if the distance tex2html_wrap_inline2985 to the next obstacle is 230cm. Again, the high probability of measuring 500cm is due to the fact that this distance represents the probability of measuring at least 500cm. The solid line in the figure represents the distribution obtained by adapting the parameters of our sensor model so as to best fit the measured data. The corresponding measured and approximated probabilities for the laser sensor are plotted in Figure 5(b).

The observed densities for all possible distances tex2html_wrap_inline2985 to an obstacle for ultrasound sensors and laser range-finder are depicted in Figure 6(a) and Figure 6(c), respectively. The approximated densities are shown in Figure 6(b) and Figure 6(d). In all figures, the distance tex2html_wrap_inline2985 is labeled ``expected distance''. The similarity between the measured and the approximated distributions shows that our sensor model yields a good approximation of the data.

  figure1189

Please note that there are further well-known types of sensor noise which are not explicitly represented in our sensor model. Among them are specular reflections or cross-talk which are often regarded as serious sources of noise in the context of ultra-sound sensors. However, these sources of sensor noise are modeled implicitly by the geometric distribution resulting from unknown obstacles.


next up previous
Next: Filtering Techniques for Dynamic Up: Metric Markov Localization for Previous: The Action Model

Dieter Fox
Fri Nov 19 14:29:33 MET 1999