You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We should make an agent car (System) that drives on Maliput.
I don't have a good name for this new System yet. We will have to figure one out.
Like SimpleCar, the System's sole input will be DrivingCommand (understood to be normalized accelerations like SimpleCar, not a velocity setpoint like car_sim_lcm).
The state will be the same as EndlessRoadCarState -- lane S/R position, S-relative heading, and speed.
One of the System's output ports will by the Isometry3d to feed into a PoseBundle; see @david-german-tri about exactly how this should work. We could also imagine a second output port with the a copy of the state, for logging / debugging purposes.
If we want to defer steering support, that's fine for a first PR. We can leave the lateral elements (R, heading) out of the state, and ignore the steering input. If we want to handle lateral motion, then EndlessRoadCar in kUser mode serves as a good example.
The system's constructor will be given a pointer to the needful Maliput elements to drive on -- possibly just a Lane* pointer.
In a first PR, the car can just stop when its Lane ends. (It only needs to operate on Dragway lanes, i.e., without any Branches.) In future PRs, it could declare a discrete event whose time is coincident with a branch choice, and just take the default branch all the time. The current Lane would thus turn into an abstract element of the declared state.
Edit: TBD -- first version without steering is the most useful, steering in R space is not relevant.
The text was updated successfully, but these errors were encountered:
We should make an agent car (System) that drives on Maliput.
I don't have a good name for this new System yet. We will have to figure one out.
Like SimpleCar, the System's sole input will be DrivingCommand (understood to be normalized accelerations like SimpleCar, not a velocity setpoint like car_sim_lcm).
The state will be the same as EndlessRoadCarState -- lane S/R position, S-relative heading, and speed.
One of the System's output ports will by the Isometry3d to feed into a PoseBundle; see @david-german-tri about exactly how this should work. We could also imagine a second output port with the a copy of the state, for logging / debugging purposes.
If we want to defer steering support, that's fine for a first PR. We can leave the lateral elements (R, heading) out of the state, and ignore the steering input. If we want to handle lateral motion, then EndlessRoadCar in kUser mode serves as a good example.
The system's constructor will be given a pointer to the needful Maliput elements to drive on -- possibly just a
Lane*
pointer.In a first PR, the car can just stop when its Lane ends. (It only needs to operate on Dragway lanes, i.e., without any Branches.) In future PRs, it could declare a discrete event whose time is coincident with a branch choice, and just take the default branch all the time. The current Lane would thus turn into an abstract element of the declared state.
Edit: TBD -- first version without steering is the most useful, steering in R space is not relevant.
The text was updated successfully, but these errors were encountered: