- Switch to openrobots2-idl
- Don't publish a reference timestamp until first valid position is computed, so that position contrlles wait for the first reference before servoing on something.
- In velocity mode, plan from the next trajectory sample to be executed, not the current one.
- Increase the default snap derivative limit (for the velocity planner)
- Add a velocity() function to reach a desired velocity in minimum time.
- Rename "reset" function to "stop", which is hopefully more clear.
- Rename set_velocity, acceleration et al. to set_velocity_limit et al.
- Fix the semantic of interrupt codels in motion activities (#147).
- Cancel any current trajectory after a new "goto" or "takeoff" (#135).
- Fix an internal race condition possibly reading a off-by-one trajectory sample.
- This is the last realease before 2.x, which will bring incompatible changes.
- Added an optional 'duration' parameter to the planning requests. Passing 0 will compute a minimum time trajectory, as before. Any other value will compute a trajectory in the specified time (unless the given duration is smaller than the minimum time, in which case the plan will abort).
- Fixed the starting point of trajectories after an abort of a previous plan.
- Fixed the behaviour of goto() that now correctly starts from the current state (and not the last state of a previous plan).
- Added a replay() function that plays back a log file or a trajectory file. The input file must have the same format as log files, but can be sampled at any frequency (samples are interpolated with the planner if the frequency is smaller than the component's one).
- Use aio(7) to write to log files, and drop log entries if the filesystem is too slow.
- Always start next trajectory from the last state reached - not the current one
- Add a set_current_state() service to reset the trajectory starting state
- Ignore velocities and acceleration when setting the current state, because the noise present in those values might generate weird trajectories.
Also available in: Atom