What's Changed

• Python cleanup by @FreyJo in #1146
• Getters for partial condensing matrices Q, R, S by @FreyJo in #1145
• Add timers: time_preparation, time_feedback by @FreyJo in #1147
• Documentation update by @josipkh in #1141
• Remove t slacks from ocp_nlp_out by @FreyJo in #1148
• Remove main mex template by @FreyJo in #1149
• Update problem formulation description by @sandmaennchen in #1150
• Fix small typo in Makefile by @josipkh in #1152
• Visualize cost-to-go in MOCP example by @FreyJo in #1153
• Add reset function in MEX OCP solver interface by @FreyJo in #1156
• Fix QP status issue by @FreyJo in #1155
• Cmake for multi-phase OCP solver by @sandmaennchen in #1157
• Simulink interface: add initialization functionalities and tests by @FreyJo in #1158
• Python interface: default for sim_method_num_stages by @sandmaennchen in #1159

Full Changelog: v0.3.3...v0.3.4

What's Changed

• Matlab race car example by @FreyJo in #1078
• Minor cleanup of chain_mass/solution_sensitivity_example.py by @dirkpr in #1080
• Fix acados cmake config by @FranekStark in #1081
• Add rti_phase as possible Simulink input by @FreyJo in #1082
• Remove local CMAKE_PREFIX_PATH adjustments by @Hs293Go in #1083
• Revert "Remove local CMAKE_PREFIX_PATH adjustments (#1083)" by @FreyJo in #1086
• Modularize CI workflow by @AkashJohnSubash in #1085
• Batch integrator and OCP solver for parallel solves by @sandmaennchen in #1087
• Globalization in minimal examples by @sandmaennchen in #1088
• Follow up to #1087 by @sandmaennchen in #1089
• Conditionally add OSQP test dependency by @FreyJo in #1091
• Create parametric closed loop example in Matlab by @FreyJo in #1092
• CMake and CI cleanup by @FreyJo in #1093
• Python 3.7 compatible interface, modified f-strings by @FreyJo in #1094
• Miscellaneous python improvements by @sandmaennchen in #1095
• Documentation update by @FreyJo in #1096
• Minor documentation update by @sandmaennchen in #1098
• Octave works by @FreyJo in #1097
• Fix cost type detection for casadi versions >= 3.6 by @sandmaennchen in #1099
• DDP and Feasibility-Projected DDP implementation by @david0oo in #1071
• AcadosSimSolver python 3.7 compatibility by @FreyJo in #1101
• improve translate_to_feasibility_problem by @FreyJo in #1100
• CI: run on Ubuntu 22.04, CasADi 3.6.5, GNU Octave 6.4.0 by @AkashJohnSubash in #1102
• Add functionality in ocp_nlp_solver_destroy by @FreyJo in #1079
• Refactor integrator experiment with Furuta pendulum by @FreyJo in #1103
• Python: check if bounds are 1-dimensional numpy arrays by @FreyJo in #1104
• fix openMP check by @hshose in #1107
• Minor cleanup in python interface by @sandmaennchen in #1108
• Multiphase slack fix by @FreyJo in #1109
• Extend create_model_with_cost_state() with slack penalties by @FreyJo in #1110
• Cleanup time-optimal python example by @sandmaennchen in #1112
• Fix merit weight initialization by @FreyJo in #1116
• Add option to log max norm of the primal step within SQP by @sandmaennchen in #1115
• Exploit symmetry when generating Hessians with CasADi SX by @FreyJo in #1117
• update qpoases version by @FreyJo in #1119
• Add soft constraints to OSQP interface by @giaf in #1120
• Fix reformulation to feasibility problem for empty symbolics by @sandmaennchen in #1121
• Batch solver: fix openmp include in templates by @sandmaennchen in #1122
• Generalize zoRO custom update function by @FreyJo in #1054
• MATLAB interface - control rates example by @josipkh in #1126
• Full soft constraints support in OSQP interface by @giaf in #1123
• Improve error messages on J matrices in Python by @FreyJo in #1127
• rename augment_model_with_polynomial_control() to reformulate_with_polynomial_control() by @FreyJo in #1129
• Feasibility Problem Reformulation enhancement by @david0oo in #1128
• Fix make build system adding DDP source, recommend CMake by @FreyJo in #1134
• Minimal example: use GN Hessian, improve warning by @FreyJo in #1135
• Setter for updating max_iter in Python by @zanellia in #1133
• Moved adaptive Levenberg marquardt calculation to common by @david0oo in #1136
• CI matlab examples by @AkashJohnSubash in #1132
• Python: improve documentation on solution sensitivities and exact hessian checks by @sandmaennchen in #1137
• Add missing external function for convex-over-nonlinear constraints by @sandmaennchen in #1139
• Windows, MATLAB - Fix timing with MinGW by @josipkh in #1142
• Option to specify partial condensing block sizes in C and Python by @FreyJo in #1143
• CMakeLists: allow specifying CMAKE_INSTALL_PREFIX directly instead of ACADOS_INSTALL_DIR by @asherikov in #1138

New Contributors

• @dirkpr made their first contribution in #1080
• @AkashJohnSubash made their first contribution in #1085
• @david0oo made their first contribution in #1071
• @hshose made their first contribution in #1107
• @asherikov made their first contribution in #1138

Full Changelog: v0.3.2...v0.3.3

What's Changed

• Allow to set cost scaling from Python by @FreyJo in #1058
• Mex add support for soft nonlinear constraints at initial shooting node by @FreyJo in #1059
• Simplify SQP loop with static functions by @FreyJo in #1057
• add idxsh_0 to layout by @FreyJo in #1060
• Fix openmp by @FreyJo in #1061
• Solution sensitivities with respect to parameters by @FreyJo in #1055
• CMake update by @Hs293Go in #1065
• Small cleanup by @FreyJo in #1064
• CSTR example improvement by @FreyJo in #1066
• Python sanity checks on idx* by @FreyJo in #1067
• Add set_params_sparse multiphase OCP by @FreyJo in #1069
• Python convenience improvements by @FreyJo in #1068
• Renaming follow up on #1055 by @FreyJo in #1070
• Improve Python unicycle example by @FreyJo in #1072
• Python interface: cleanup solver by @sandmaennchen in #1073
• Matlab Interface: Add ztraj to sfunction output by @leonardocecchin in #1074

Full Changelog: v0.3.1...v0.3.2

What's Changed

• fix Matlab with MSVC by @FreyJo in #1036
• Support slack penalties for initial stage in MEX interface by @FreyJo in #1037
• Split condensing and regularization into preparation and feedback by @FreyJo in #1034
• Work on core for speedups by @FreyJo in #888
• Fix slack function update in CONL, NLS module by @FreyJo in #1039
• Cleanup RTI, fix timings and comments by @FreyJo in #1038
• Cleanup SQP submodules - only use single evaluation point at a time by @FreyJo in #1040
• Fix a few typos regarding the least squares cost formulation (MATLAB examples, Python docs) by @josipkh in #1044
• Minor C changes by @FreyJo in #1045
• Implement advanced-step real-time iteration: AS-RTI by @FreyJo in #1046
• Fix DAQP for Matlab on Windows by @FreyJo in #1043
• Efficient factorization of diagonal outer Hessian in convex-over-nonlinear and nonlinear least-squares costs by @sandmaennchen in #1024
• Update MATLAB examples with general initial input constraints by @josipkh in #1048
• Fix solution sensitivities in Matlab by @FreyJo in #1050
• Add DS1104 support by @FreyJo in #1052
• Furuta example by @sandmaennchen in #1053

Breaking:

• Calling ocp_nlp_precompute is now required when creating an OCP solver from C, this is done in the C files generated by the template based interfaces (Python, Matlab and Octave). It might affect users that use an old / self written C file to set up their OCP solver. See: 37e17d3

Full Changelog: v0.3.0...v0.3.1

What's Changed

• docs: update Python requirement by @FreyJo in #997
• Matlab: LD_RUN_PATH, add getters, fix examples, remove outdated files by @FreyJo in #996
• Separate nonlinear constraints into path and initial stage constraints by @FreyJo in #994
• Some interface refactoring by @FreyJo in #998
• Fix solver template, simplify external function generation, formatting by @FreyJo in #1000
• Fix bugs with slack dimensions Python interface by @FreyJo in #1001
• Matlab: allow to create Simulink block from integrator by @FreyJo in #1003
• Add support for multi phase OCP in Python by @FreyJo in #1004
• Fix acados_timer issue by removing MEASURE_TIMINGS flag by @FreyJo in #1005
• Python 3.7 compatibility by @FreyJo in #1006
• Fix nonlinear_chain_example, removing nm6 by @FreyJo in #1008
• CodeQL: dont complain about commented code by @FreyJo in #1010
• Consistent integrator timings by @FreyJo in #1011
• extend AcadosMultiphaseOcp by @FreyJo in #1007
• Add a linear MPC example in MATLAB by @josipkh in #1012
• Documentation: fixes and updates by @FreyJo in #1013
• Fix regularization example and build with CMake by @FreyJo in #1015
• Update documentation, fix some typos by @josipkh in #1016
• Minor fixes by @FreyJo in #1018
• Update HPIPM by @FreyJo in #1020
• Update penalty utilities by @FreyJo in #1021
• Fix multiphase nonlinear constraints by @FreyJo in #1023
• Matlab fixes by @FreyJo in #1026
• Matlab examples: make calling functions more explicit by @FreyJo in #1027
• Implement fixed_hess option to not recompute the Hessian if it is constant by @FreyJo in #1028
• AcadosOcpSolver: make ocp consistent if generate==False by @FreyJo in #1030
• Update osqp by @FranekStark in #1033

Breaking:

• Matlab & Python: nonlinear constraints h and phi(r) are by default only enforced at nodes 1 to N. Such constraints can still be enforced at node 0 by setting the appropriate new fields, #994

New Contributors

• @josipkh made their first contribution in #1012
• @FranekStark made their first contribution in #1033

Full Changelog: v0.2.6...v0.3.0

What's Changed

• fix CodeQL warnings by @FreyJo in #984
• fix condition for checking number of stages in detect_dims_ocp.m by @Ajin2305 in #989
• Codeql review by @FreyJo in #988
• Rework CasADi requirements in MEX interface by @FreyJo in #991
• Fix get_optimal_value_gradient, add getter for p from HPIPM by @FreyJo in #993
• Windows python interface by @asparc in #968

New Contributors

• @Ajin2305 made their first contribution in #989

Full Changelog: v0.2.5...v0.2.6

What's Changed

• update BLASFEO and HPIPM by @FreyJo in #969
• Fix typo in minimal_example_closed_loop.py by @Federico-PizarroBejarano in #970
• Fix dependency handling in CMake config file by @Hs293Go in #971
• Fix sfun sources if hessian is not exact by @asparc in #974
• update ROADMAP by @sandmaennchen in #964
• Work on solution sensitivities by @FreyJo in #975
• Implement cost integration via IRK for Convex-over-nonlinear cost by @FreyJo in #976
• Add time in IRK, allow to cost integration with time dependent function by @FreyJo in #977
• Documentation fixes by @FreyJo in #979
• Formulate constraints as L2 and Huber penalties in Python by @FreyJo in #980

New Contributors

• @Federico-PizarroBejarano made their first contribution in #970
• @Hs293Go made their first contribution in #971

Full Changelog: v0.2.4...v0.2.5

What's Changed

• Zoro followup by @FreyJo in #960
• GNRK: Gauss-Newton Runge-Kutta: IRK with nonlinear-least squares by @FreyJo in #698
• Python get_stats returns float or int instead of single element numpy array and fix cython for custom path by @aghezz1 in #961
• python interface: get the gradient of the optimal value function by @sandmaennchen in #963
• Cleanup by @FreyJo in #965
• refactored optionally verbose system calls in python interface by @tudoroancea in #962

BREAKING: change return types of get_stats in Python

Full Changelog: v0.2.3...v0.2.4

What's Changed

• Minor update to python interface docs, remove docs folder in acados_template by @sandmaennchen in #943
• minor changes of zoRO examples by @yf-gao in #946
• Simulink with custom update function by @FreyJo in #948
• Core: Allow changing all constraint bounds between preparation and feedback phase by @FreyJo in #949
• Python x0 shape check by @FreyJo in #950
• MATLAB: use template-based code to interface acados sim solver by @FreyJo in #944
• Fix free external function in templated sim solver by @sandmaennchen in #951
• MEX: remove pointers to external functions in Matlab by @mohammed-hababeh in #952
• Use PROJECT_SOURCE_DIR instead of CMAKE_SOURCE_DIR to allow for better submodulization by @apozharski in #953
• Update problem_formulation_ocp_mex.tex with J_sbx^e by @jnz in #954
• Refactor examples by @FreyJo in #955
• Minimal closed-loop example using RTI by @sandmaennchen in #958

New Contributors

• @mohammed-hababeh made their first contribution in #952
• @apozharski made their first contribution in #953
• @jnz made their first contribution in #954

Full Changelog: v0.2.2...v0.2.3

What's Changed

• Update License headers by @FreyJo in #935
• Matlab full template based OCP solver by @yzuuang in #934
• Fast implementation of the zoRO algorithm by @yf-gao in #938
• NLS cost module: Change W_changed flag if cost matrix was changed by @lukasfro in #939
• Example to create Simulink wrapper from Python by @FreyJo in #940
• license follow up: fix trailing spaces by @FreyJo in #941

Breaking changes:

• Simulink interface: After #934 the Matlab OCP solver interface is fully template based. Thus, there is no need anymore to call ocp.generate_c_code(simulink_opts);. This is now done within the creation of the Matlab OCP solver object.
  The simulink_opts, have to be provided to the constructor of the ocp solver instead, i.e. with the optional argument simulink_opts:

ocp = acados_ocp(ocp_model, ocp_opts, simulink_opts);

For details, see #934

New Contributors

• @yf-gao made their first contribution in #938

Full Changelog: v0.2.1...v0.2.2