From 06ae48469025690430e8b9693bee9eebd06ff85d Mon Sep 17 00:00:00 2001 From: mvoellmy Date: Fri, 12 Jun 2015 20:32:15 +0200 Subject: [PATCH] if fmincon did not converge the angle computations are not executed, additionally if this happens for more than 3 times matching gets reinitialized commented everything within joint_callback out stair model size has been updated changed dx reinitialization to only happen if the stair is acctually curved --- CHANGELOG.rst | 16 +++++++-- launch/angle.launch | 4 ++- matlab/12-Jun-2015 17:56:32.fig | Bin 0 -> 11319 bytes matlab/12-Jun-2015 19:42:37.fig | Bin 0 -> 16210 bytes matlab/stairparam.m | 2 +- matlab/stairparam.mat | Bin 2918 -> 7888 bytes src/angle.cpp | 56 +++++++++++++++++++------------- src/angle.h | 6 +--- src/stair_model.cpp | 2 +- 9 files changed, 53 insertions(+), 33 deletions(-) create mode 100644 matlab/12-Jun-2015 17:56:32.fig create mode 100644 matlab/12-Jun-2015 19:42:37.fig diff --git a/CHANGELOG.rst b/CHANGELOG.rst index fbb4b6d..35e0754 100644 --- a/CHANGELOG.rst +++ b/CHANGELOG.rst @@ -5,31 +5,43 @@ Changelog for package scalaser TODO ------------------ * add kp to plot +* change matlab plots to change data setXdata, setYdata +* define beta threshold in launch file +* add folder for figures KNOWN BUGS ------------------ * Fix inconsistent performance of the angle node. It ranges from 1 Hz to 4 Hz. If a rviz is started the frequency takes a hit, however once it's closed again the node wont performe as well as before. +* joint_states influence towards the initial guesses causes a big mess for the stair_model NICE TO HAVE ------------------ +* add radius-tracking +* add 3D model for curved stairs +* field of view initialization +* add field of view in rviz * clean up v_r_1 and dx_1 in angle.cpp * add initialization for start and end of stairs to determine when to stop the stair model * add timestamp to published messages * make fov_s, fov_d dependant of the lambda position received from the MyRIO * make stair growth dependant of fov_s -* define beta threshold in launch file * make the velocity_forward sent to the MyRio adjustable with dynamic reconfigure * (use IMU values to set guess for beta threshold value -> not sure if needed, since the wheelchair never comes close to a beta=~24°) * (change the initialization of v0 from within the angle constructor to the matching constructor since startvalues for both sides are identical) * (change tf static to tf2 static) +0.1.00 (2015-06-12) +------------------- +* if fmincon did not converge the angle computations are not executed, additionally if this happens for more than 3 times matching gets reinitialized +* commented everything within joint_callback out +* stair model size has been updated +* changed dx reinitialization to only happen if the stair is acctually curved 0.0.14 (2015-06-10) ------------------- * added filter for scan data to be between 0.3 and 1 m * changed beta threshold -* 0.0.14 (2015-06-08) ------------------- diff --git a/launch/angle.launch b/launch/angle.launch index e754ac1..fa0a349 100644 --- a/launch/angle.launch +++ b/launch/angle.launch @@ -2,7 +2,7 @@ - + @@ -11,4 +11,6 @@ + + \ No newline at end of file diff --git a/matlab/12-Jun-2015 17:56:32.fig b/matlab/12-Jun-2015 17:56:32.fig new file mode 100644 index 0000000000000000000000000000000000000000..14124d1e20ab0ef849658eb57ca461d74903f7be GIT binary patch literal 11319 zcma)iRZtv2(=85x;3SaX9)i0q3j~J%AwaO;7Tlf1HMm2N4H5_hcXtUMAh77-E{iO> zEPL~PfBm=WzTK{wsh+N$nx6B}b$Y7RWp&=mzGD#P;bTyj)#0|WbF}7S_~2;nY2)JV zB*CEcUPE72M2L$)&fVJF)7px`#aV(u(cO+g)ytWIUx0yMR6YBkofCvDZ8{Mb;Cnh(bvXPUj4Rhvn>?h