Dum loquimur, fugerit invida aetas: carpe diem, quam minimum credula postero. (Orazio, Odi, I, 11, 7-8)
oRatio is an Integrated Logic and Constraint based solver which takes inspiration from both Logic Programming (LP) and Constraint Programming (CP).
The oRatio system is invoked with a list of command line arguments representing the locations of the required compilation units (e.g. domain models, problem instances or plan adaptations) and the desired output file like in the following:
oRatio cu_0.rddl cu_1.rddl ... cu_n sol.json
As an example, the following code invokes the oRatio solver using the domain model defined in examples/blocks/blocks_domain.rddl
, the problem instance specified in examples/blocks/blocks_problem_10.rddl
and, once found a solution, writes it in the solution.json
file:
oRatio examples/blocks/blocks_domain.rddl examples/blocks/blocks_problem_10.rddl solution.json
Further information about the RIDDLE language, used for specifying the input files, can be found in the corresponding wiki page.
The basic requirements for building oRatio are:
The easiest way to install the building requirements on Ubuntu is as follows
sudo apt-get install build-essential
sudo apt-get install cmake
once the building requirements are installed, move to a desired folder and clone the oRatio repository
git clone https://github.com/pstlab/oRatio
finally, build oRatio
mkdir build
cd build
cmake ..
make
The easiest way to install the building requirements on OS X consists in downloading and installing the Command Line Tools and CMake. Once the building requirements are installed, open a terminal, move to a desired folder and clone the oRatio repository
git clone https://github.com/pstlab/oRatio
finally, build oRatio
mkdir build
cd build
cmake ..
make
The easiest way to compile oRatio on Windows is through Visual Studio. Download and install Visual Studio, download Git and CMake. Start a Visual Studio Command prompt, move to a desired folder and clone the oRatio repository
git clone https://github.com/pstlab/oRatio
finally, build oRatio
mkdir build
cd build
cmake -G "NMake Makefiles" ..
nmake
The CMake building script, by default, generates a 'Debug' target resulting in a code that, when executed, performs several checks and is not optimized. A different target can be choosen either from the CMake gui, by setting the CMAKE_BUILD_TYPE
parameter to the desired target, or from the command line, by providing CMake the same parameter through the -D
option.
Available targets are:
- Debug, emits debug information and do not performs optimization
- Release, omits debug information and performs optimization
- MinSizeRel, optimizes for smallest binary
- RelWithDebInfo, emits debug information and performs optimization
As an example, an optimized target can be generated through the following command
mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make
Integrating oRatio into a ROS environment is a straightforward task! You just need clone this repository into a catkin workspace and compile the workspace. That's all!
oRatio contains a catkin package which, through the implementation of some ROS services and messages, manages the communication to and from other ROS nodes. The package provides ROS services for the creation of reasoners, problem solving, their execution and their adaptation during execution. If on the one hand oRatio deals with the management of the reasoners, on the other hand some ROS services must be implemented by the invoking modules, so as to allow the execution of the dispatched activities.
More details about how to interface with oRatio in a ROS environment can be found here.