1.5 Version history

version 1.32

(2013/12/12)

OpenWatcom support is dropped.

Upgraded the matrix library to NEWMAT11 (beta) November 2008

Code clean up dealing with some warnings.

Call to Gnuplot under Windows is now using pipes.

Removed CVS keywords tags.

inv_kin immobile joint index bug corrected (thanks to Matteo Malosio).

version 1.31

(2006/12/14)

The project can now use CMake or Bakefile for automated makefile generation. In future releases, hand made makefiles and project files will be replaced by the output of CMake or Bakefile.

Corrected bug in irotk (reported by Chris Lightcap).

version 1.30

(2006/08/17)

Upgraded the matrix library to NEWMAT11 (beta) April 2006 enabling compilation under GNU g++ 4.1.x.

version 1.29

(2006/05/19)

OpenWatcom support is (temporally) suspended. Fixed gear ratio bug for viscous friction (reported by Carmine Lia). Fix set_q, set_qp bug in xdot (reported by Philip Gruebele)

The following changes have been contributed by Etienne Lachance

• “Clean up” of some header files.
• Member functions add and select are now in template form.
• Using Boost shared pointers in gnugraph.
• The inverse kinematics function (inv_kin) should return the solution without changing the robot position (reported by J.D. Yamokoski).
• Functions Rhino_DH, Puma_DH, Schilling_DH, Rhino_mDH, Puma_mDH and Schilling_mDH use const Robot_basic reference instead of const Robot_basic pointer.
• Prevent exceptions from leaving Robot_basic destructor.
• Catch exception by reference instead of by value.

version 1.28

(2005/12/07)

The following changes have been contributed by Etienne Lachance

• Removing unnecessary copy constructor and the assignment operator (operator=) in many classes.
• In the Quaternion class, the operator* and operator/ are now non-member functions when one of the operand is a real, it now supports q2 = c * q1 and ¿ q2 = q1 * c

version 1.27

(2005/10/11)

It is now possible to turn off warning messages in the Config class.

version 1.26

(2005/07/05)

• New Class Stewart contributed by Samuel Belanger (intergated by Etienne Lachance and Richard Gourdeau): new files stewart.h and stewart.cpp and modified bench.cpp.
• Fixed max() bug for VC++ 6.0 (utils.cpp).
• Typos in Doxygen documentation.

version 1.25

(2005/06/13) Fixed catch(bad_alloc) in constructors.

The following changes have been contributed by Etienne Lachance

• The desired joint acceleration was missing in the computed torque method (bug reported by Carmine Lia).
• Added missing file message in trajectory.cpp

The following changes have been contributed by Carmine Lia

• Added defined(__MINGW32__) for temp files in gnugraph.cpp.
• Added pinv in utils.cpp.

version 1.24

(2005/03/18)

The following changes have been contributed by Brian Galardo, Jean-Pascal Joary, Etienne Lachance:

• Added member functions Robot::inv_schilling,

  mRobot::inv_schilling and mRobot_min_para::inv_schilling for the Schilling Titan II robot arm,

• Fixed previous bug on Rhino and Puma inverse kinematics.

by Etienne Lachance:

• Some “clean-up” in the config.h and config.cpp files,

and by Stephen Webb :

• minor bug in constructor Robot_basic(const Robot_basic & x).

version 1.23

(2004/09/18)

The following change has been contributed by Etienne Lachance:

• Configuration files can use degrees for the angles with the option angle_in_degree set to 1.

version 1.22

(2004/09/10)

The following change has been contributed by Etienne Lachance:

• In config.cpp: parameter value can now contain space and fixed print member function.

Carl Glen Henshaw provided a makefile for MAC OS X.

version 1.21

(2004/08/16)

The following changes have been contributed by Etienne Lachance

• Fixed some missing use_namespace #define.
• Merge all select_* and add_* functions into overloaded select() and add() functions.
• made gnuplot.cpp and config.cpp independent of robot.h and utils.h.
• New constructors for Robot and mRobot based on input matrices (this change is NOT backward compatible)

The following changes have been contributed by Ethan Tira-Thompson

• Supports for Link::immobile flag so jacobians and deltas are 0 for immobile joints.
• Jacobians will only contain entries for mobile joints - otherwise NaNs result in later processing.
• Added parameters to jacobian functions to generate for frames other than the end effector.
• Can now do inverse kinematics for frames other than end effector.
• Tolerance in inv_kin based on USING_FLOAT from newmat’s include.h

version 1.20

(2004/07/02)

The following changes have been contributed by Ethan Tira-Thompson

• Added support for newmat’s use_namespace #define, using ROBOOP namespace.
• Fixed some problem using float as Real type.

The following changes have been contributed by Etienne Lachance

• Added the following class: Dynamics, Trajectory_Select,

  Proportional_Derivative and Control_Select.

• Added a new demo program, call demo_2dof_pd. This new demo program shows how to use the class mentioned above.
• Protection added on input vector of the trans function.
• Added a joint_offset logic. This idea has been proposed by Ethan Tira-Thompson.
• Added Doxygen documentation.
• Replace files impedance.* by controller.*.

version 1.19

(2004/05/12) Upgraded the matrix library from NEWMAT10 to NEWMAT11 (beta). Visual C++ .NET and Borland C++ Builder 6 compilers are now supported. Updated documentation.

version 1.18

(2004/05/05) ROBOOP is relicensed to the GNU Lesser General Public License. Updated documentation.

The following changes have been contributed by Vincent Drolet and Etienne Lachance:

• Added the following members function in class Robot: inv_kin_rhino, inv_kin_puma and robotType_inv_kin.

version 1.17

(2004/04/02) Numerous warning messages were corrected under VC++. Updated documentation.

The following changes have been contributed by Etienne Lachance:

• Added class Impedance which implements the impedance controller.
• Added function perturb_robot.
• Added class Resolve_acc which implements the resolve rate acceleration position controller.
• Added class Computed_torque_method which implements the computed torque method position controller.
• Class Config can now write data into a configuration file.
• Fixed bugs in Quaternion class member functions: exponential and logarithm.
• Added Quaternion class member function power.
• Added the following Quaternion class non member functions: Omega, Slerp, Slerp_prime, Squad and Squad_prime.
• Provided Spl_Quaternion class to generate quaternions cubic splines.
• Added class Spl_Cubic to generate cubic splines.
• Added class Spl_path to generate 3D cubic splines.
• Provided CLIK class for closed loop inverse kinematics.
• Added member functions G and C in all robot classes.

version 1.16

(2003/09/24) The OpenWatcom C++ compiler is now supported. Updated documentation.

version 1.15

(2003/06/18) The following changes have been contributed by Etienne Lachance:

• Updated documentation.
• Definitions in file gnugraph.cpp are now in gnugraph.h.
• Class Plot2d, GNUcurve are now using STL string instead of char*.
• Added member functions jacobian_dot() and jacobian_DLS_inv() in all robot classes.
• Added class Config to read configuration file.
• Replaced Robot_basic(const char *filename) by Robot_basic(const string & filename). The new constructor uses the class Config.
• Provided Plot_file class to generate graphics from a data file.
• Added the following Quaternion class member functions: exponential, logarithm, dot_product, dot, E.
• Fixed bugs in IO_matrix_file class.
• Developed linearized equations for modified DH notations. The equations are implemented in dq_torque, dqp_torque, dtau_dq and dtau_dqp.
• Added examples in demo.cpp related to IO_matrix_file, Plot_file and Config.

version 1.14

(2003/04/17) Updated documentation. The Watcom compiler is no longer supported (problems with STL and streams). The following changes have been contributed by Etienne Lachance:

• The classes RobotMotor and mRobotMotor no longer exist and are now integrated in the Robot and mRobot classes.
• The Robot and mRobot classes are now derived from the Robot_basic virtual class.
• Removed class mlink. DH and modified DH parameters are now included in link.
• Added kine_pd().
• Created a new torque member function that allowed to have load on last link.
• Fixed bug in modified DH dynamics.
• Added a class Quaternion.
• Added the program rtest to compare results with Peter Corke MATLAB toolbox.
• Added member function set_plot2d to generate plots using the Plot2d class.
• Added utility class IO_matrix_file dealing with data files (not documented yet).

version 1.13

(2002/08/09) Moved the arrays of ColumnVector to the constructors for the dynamics and linearized dynamics for a ≈ 10% gain in speed (thanks to Etienne Lachance for the suggestion). Added the mRobot and mRobotMotor classes using the modified Denavit-Hartenberg notation. Updated documentation.

version 1.12

(2002/02/04) Upgraded the matrix library from NEWMAT09 to NEWMAT10.

version 1.11

(2001/06/06) Fixed bugs for prismatic joints in the dynamics routines (reported by Hassan Abedi). Updated documentation.

version 1.10

(2001/04/30) Changed the license to GNU General Public License. Workspace for MS Visual C++ 6.0. New makefiles using implicit rules. New class RobotMotor that includes motors parameters (rotor inertia, gear ratio and friction coefficients). Updated documentation.

version 1.09

(98/09/27) Makefile for MS Visual C++ 6.0.

version 1.08

(98/06/1) Changes to robot.cpp and robot.h to avoid the warning messages:

initialization of non-const reference ‘*’ from rvalue ‘*’

Fixed function ieulzxz in homogen.cpp thanks to Kilian Pohl.

version 1.07

(98/05/12) The bench.cpp program is more portable. Simpler makefile for Borland C++. New targets in makefiles (clean and veryclean). Removed the CVS Log tags from the sources. Compiler option -O now works under gcc 2.7.2 thanks to the new newmat.h provided by Robert Davies.

version 1.06

(97/11/21) The function inv_kin modified to use the Jacobian by default in the iterative procedure (≈ 1.8× faster). Updated documentation.

version 1.05

(97/11/17) Added make file for GNU G++ under Windows 95/NT using Cygnus GNU-Win32 compiler. Added optimization flags under GNU G++. Updated documentation.

version 1.04

(97/11/14) Added make file for GNU G++ and graphic support through gnuplot (2d plots). Updated documentation.

version 1.03

(97/11/01) Added adaptive step size integration. Changes to the documentation.

version 1.02

(97/10/21) Upgraded the matrix library from NEWMAT08A to NEWMAT09. New directory structure : newmat08 is replaced by newmat. Conditional compilation of delete [] for pre 2.1 C++ compilers has been removed since NEWMAT09 no longer supports these compilers. Minor changes to the documentation.

version 1.01

(97/01/17) Conditional compilation of delete [] for pre 2.1 C++ compilers. Changes to the documentation.

version 1.0

(96/12/15) First public release of the package.