From dd90d3c98f465b8b772ff2d195e016f94f42eea7 Mon Sep 17 00:00:00 2001 From: matlabbe Date: Mon, 10 Jul 2017 13:35:34 -0400 Subject: [PATCH] Updated Kinect mapping (markdown) --- Kinect-mapping.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Kinect-mapping.md b/Kinect-mapping.md index 237c42e..9f471f9 100644 --- a/Kinect-mapping.md +++ b/Kinect-mapping.md @@ -67,7 +67,7 @@ Tutorial about mapping an environment using a hand-held Kinect. * When the background turns RED, it means that odometry cannot be computed, i.e. the odometry is lost and the mapping cannot continue. These events would cause RED screens: 1. Featureless environments: white walls, uniform texture, dark areas, etc. Without enough features, odometry cannot be computed. Odometry quality would be over 100 for environments with enough visual features. - 2. "Empty space" environments: For sensors like the Kinect, good features are within around 0.4 to 4 meters of the sensor. By default, features farther than 4 meters are not used. For example, if you point the sensor over a clear area where there are no objects under 4 meters, there are good chances that the odometry will lose tracking. + 2. "Empty space" environments: For sensors like the Kinect, good features are within around 0.4 to 4 meters of the sensor. For example, if you point the sensor over a clear area where there are no objects under 4 meters, there are good chances that the odometry will lose tracking. 3. Too fast camera motion: If the camera is moved too fast, there will be less features to match between successive frames. 4. A misalignment between the depth and RGB images: If the depth of the visual features is wrong, this would have a negative impact on the odometry. Make sure that depth images are correctly registered to RGB images like [here](https://github.com/introlab/rtabmap/wiki/Installation#depth-not-registered-to-rgb).