This is a readme file to explain the data set and a SLAM program. The algorimth details in generating such a data set can be found in Zhao, H. Chiba, M., Shibasaki, R., Shao, X., Cui, J., Zha, H., A laser scanner based approach towards driving safety and traffic data collection, IEEE Trans. Intelligent Transportation System, vol.10, no.3, 534-546, 2009. If you use the data or viewer, you are obligated to cite our research in your report. For any question, please contact: zhaohj@cis.pku.edu.cn [Data Sets] The data set is generated on a run around the Science Buildings of PKU. It contains a data file in LMS format. [Configuration.ini for SlamTracking.exe] When starting the SlamTracking.exe, it will read the configuration.ini at the same directory for program setting. Parameters in configuration.ini are defined as follows. 1. #SEVERNM1=name MODE #MODE: _SEND_VOI=0, _SEND_LOC=1, _SEND_GLO=2 SEVERNM1=npc4 1 //Explanation: //If a SEVERNM is defined, SlamTracking results, such as vehicle pose, moving objects, laser data will be sent to the server with the PC name. The data might be in sensor's local coordinate system, if mode is set to 1, or global coordinate system, if mode is set to 2. No data will be sent if mode is set to 0. 2. [Sensor Info] #CLIENT=NAME ANGRNG ANGRES UNIT L1=FM 180.0 0.5 100 G1=G //Explanation: //The program will read the data input through network according to definition. //L means laser data, followed by the ID of the sensor. The sensor data with other sensor ID will be ignored. If no or wrong laser data is read, check the sensor ID!!! //G means GPS/IMU data. The ID followed are not used now. 3. [Calib Info] #CLIENT=RAW PITCH YAW X Y Z L1=0 0 0 0 0 0 G1=0 0 0 0 0 0 //Explanation: //Sensor geometries with respect to vehicle frame. 4. [Map Data] #MAP=FILENAME MAP=C:\1st-run //Explanation: //Not a requirement. //If a map is used to assist localization, define this field. //The map data contains two pairwised files, C:\1st-run.master and C:\1st-run.raw, in this example. 5. [Simu Data] #CLIENT=FILENAME L1=C:\1st-run.lms5 //Explanation: //If in an online mode, delete these fields. //These fields are only for offline processing mode. //The program will read the LMS data, simulate the procedure of online sensor data read. 6. [Slam Option] ONLINELOOPCLOSURE=1 #True=1, False=0 GPSASSIST=0 #True=1, False=0 //Explanation: //Validate or invalidate the online loop closure and GPS assistant. [Usage SlamTracking.exe] 1. Start SlamTracking.exe 2. Run a. If the trajectory data of moving objects should be save, do Output->Output Trajectory b. If the initial coordinate system of the sensor data should be corrected, do Option->Intervention If the sno is 0, the data will be transformed as a rigid body one. If the sno is a non-zero value, the data will be corrected as a non-rigid body one, so that the data of sno will keep as it is, the other data will be transformed gradually, so that the data of eno will meet the destination frame. You can choose the "Auto Reg", so that to automatedly match the current scan data (eno) with the map, defined in Map Data. You can choose the "Manu Reg", put focus on the main window, use arrow keys for shift, Home and End keys for rotation, Ctrl+ for fine tuning. After you finished manual adjustment, push the button "Manu Reg" again, so that to finalize or abort the process. You can choose the "Loop Closure", so that to conduct a matching between the current scan data (eno) with the past run. If a matching is taken, transform the data as a non-rigid body one. c. If a loop closure should be manually operated, do Option->Intervention see above 3. Save Results Output->BG map, save a bitmap of stationary objects in *.master and *.raw format; Output->Navi Info, save vehicle pose in *.nav format; Output->Dsx, save laser scan as well as vehicle pose and lables in extended DS format. 4. The processing results are broadcast in an online mode through network.