Method, medium, and system estimating pose of mobile robots

A movement system, the system comprising:

• a particle filter unit to determine poses and weights of current particles by applying a detected pose variation to previously identified particles, with the detected pose variation being detected by a sensor to detect a pose variation in a pose of a mobile element;
• an evolution unit to update the determined poses and weights of the current particles by applying an evolution algorithm to the determined poses and weights of the current particles to estimate a current pose of the mobile element;
• a camera to capture an image of an outside of the mobile element;
• and
• a feature point processing unit to extract a least one feature point from the captured image and to create a map by processing the extracted feature point and the determined poses,
• wherein the extracted feature point is scale-invariant and rotation-invariant from the captured image,
• wherein the evolution unit comprises a mutation applicator to select a parent sample from among a plurality of evolution samples, generate a child sample by applying random noise to the selected parent sample, and select one of the selected parent sample and the child sample that has a higher weight value, with the evolution samples being obtained from one of a plurality of available particles through random sampling, and
• wherein the selected parent sample is an evolution sample whose weight value exceeds a predefined threshold.

A mobile movement controlling method, the method comprising:

• determining poses and weights of current particles by applying a detected pose variation to previously identified particles, with the detected pose variation being a pose variation in a pose of a mobile element;
• updating and outputting the determined poses and weights of the current particles by applying an evolution algorithm to the determined poses and weights of the current particles to estimate a current pose of the mobile element;
• capturing an outside image of the mobile element;
• and
• extracting at least one feature point from the captured image for obtaining an observed pose and creating a map by processing the extracted feature point and the determined poses, with the processing of the extracted feature point including determining whether one of plural currently calculated gray gradient vectors for a current feature point match a previously calculated gray gradient vector for a previously extracted feature point,
• wherein the extracted feature point is scale-invariant and rotation-invariant from the captured image.

A mobile movement controlling method, the method comprising:

• determining poses and weights of current particles by applying a detected pose variation to previously identified particles, with the detected pose variation being a pose variation in a pose of a mobile element;
• updating and outputting the determined poses and weights of the current particles by applying an evolution algorithm to the determined poses and weights of the current particles to estimate a current pose of the mobile element;
• capturing an outside image of the mobile element;
• and
• extracting at least one feature point from the captured image for obtaining an observed pose and creating a map by processing the extracted feature point and the determined poses,
• wherein the extracted feature point is scale-invariant and rotation-invariant from the captured image,
• wherein the updating comprises:
• selecting a parent sample from among a plurality of evolution samples that are obtained from one of a plurality of available particles through random sampling;
• generating a child sample by applying random noise to the selected parent sample;
• and
• selecting one of the selected parent sample and the child sample that has a higher weight value, and
• wherein the selected parent sample is an evolution sample whose weight value exceeds a predefined threshold.

A movement system, the system comprising:

• a particle filter unit to determine poses and weights of current particles by applying a detected pose variation to previously identified particles, with the detected pose variation being detected by a sensor to detect a pose variation in a pose of a mobile element;
• an evolution unit to update the determined poses and weights of the current particles by applying an evolution algorithm to the determined poses and weights of the current particles to estimate a current pose of the mobile element;
• a camera to capture an image of an outside of the mobile element;
• and
• a feature point processing unit to extract a least one feature point from the captured image and to create a map by processing the extracted feature point and the determined poses,
• wherein the extracted feature point is scale-invariant and rotation-invariant from the captured image,
• wherein the detected pose variation is a pose estimation, as a probability distribution, of the mobile element based on measured movement, by the sensor, of the mobile robot, and
• wherein the probability distribution is based at least on a conditional probability p(xt|xt−1, ut−1), where xt−1 indicates a pose at a time (t−1), ut−1 indicates data measured by the sensor, and xt indicates a pose at a current time t.

A mobile movement controlling method, the method comprising:

• determining poses and weights of current particles by applying a detected pose variation to previously identified particles, with the detected pose variation being a pose variation in a pose of a mobile element;
• updating and outputting the determined poses and weights of the current particles by applying an evolution algorithm to the determined poses and weights of the current particles to estimate a current pose of the mobile element;
• capturing an outside image of the mobile element;
• and
• extracting at least one feature point from the captured image for obtaining an observed pose and creating a map by processing the extracted feature point and the determined poses,
• wherein the extracted feature point is scale-invariant and rotation-invariant from the captured image,
• wherein the detected pose variation is a pose estimation, as a probability distribution, of the mobile element based on measured movement, by a sensor, of the mobile robot, and
• wherein the probability distribution is based at least on a conditional probability p(xt|xt−1, ut−1), where xt−1 indicates a pose at a time (t−1), ut−1 indicates data measured by sensor, and xt indicates a pose at a current time t.