ST-RRT*: a bidirectional, time-optimal planner for planning in space-time.
Multi-level planners: Planning algorithms which can exploit multiple levels of abstractions.
Rapidly-exploring Random Quotient space Trees (QRRT): A generalization of RRT to plan on different abstraction levels.
QRRT*: An asymptotically optimal version of QRRT.
Quotient-Space Roadmap Planner (QMP): A generalization of PRM to plan on different abstraction levels.
QMP*: An asymptotically optimal version of QMP.
AIT* has been significantly refactored.
SST now uses the intermediate solution callback to report new solutions.
The kinodynamic version of SST (ompl::control::SST) now supports optimization objectives.
New topological state spaces have been added: a torus, a sphere, a Möbius strip, and a Klein bottle.
Updated docker images to Ubuntu Jammy.
Several fixes for Python bindings.
Bug fixes.
1.5.0
3 years ago
A C++14 compiler is now required. The minimum version of CMake required is now 3.5 and the minimum version of Boost supported is now 1.58.
All development now takes place on Github. This used to be a git mirror of the mercurial repository on BitBucket, but since BitBucket is phasing out mercurial support the GitHub repo is now the main repo. All the old issues have been migrated to GitHub.
XXL: a probabilistically complete sampling-based algorithm designed to plan the motions of high-dimensional mobile manipulators and related platforms.
ABIT*: an extension to BIT* that uses advanced graph-search techniques to find initial solutions faster.
AIT*: an anytime asymptotically optimal algorithm that simultaneously estimates and exploits problem-specific heuristics.
Quotient-Space RRT: a generalization of RRT to plan on different abstraction levels. The abstraction levels are represented by quotient-spaces.
Taskspace RRT: a variant of RRT where exploration is guided by the task space.
RLRT and BiRLRT: basic tree-based planners without any sophistic heuristics to guide the exploration, useful as a baseline for comparison against other tree-based planners.
PRM, PRM*, LazyPRM, and LazyPRM* can now be initialized with an ompl::base::PlannerData instance (the generic way to represent roadmaps/trees in OMPL). This means that you seed these planners with data from a previous run from any other planner. Using the ompl::base::PlannerDataStorage functionality, this data can be saved to or loaded from disk.
Added support for deterministic sampling. Halton sampling is included, other deterministic sampling methods can be added.
Added a new PlannerTerminationCondition called CostConvergenceTerminationCondition, which can be used to terminate asymptotically (near-)optimal planners based on convergence.
Clean up ompl_benchmark_script.py for Python 3.
Updated PlannerArena again to work with latest versions of R dependencies.