Comments (4)
The trajectory proposals of PDM-Closed
are a pairwise combination of 3 lateral centerline offsets and 5 IDM policies. For the centerline offsets we have Β±1m and 0m. Therefore, itβs perfectly normal that the trajectory of PDM-Closed
is not consistent with the centerline and runs parallel to the centerline (also see here in code). There might also be inconsistencies due to precision or interpolation errors, which should be negligible.
Note that PDM-Hybrid
does use PDM-Offset
for long-term correction. PDM-Offset
follows the architecture and input of PDM-Open
but additionally receives the output trajectory of PDM-Closed
as input that includes the centerline offset (see here in code).
The PDM-Hybrid
output often looks disturbed, especially if PDM-Offset
is out-of-distribution during closed-loop simulation. Could you share the scenario token of the third image? I might have a look.
from tuplan_garage.
@DanielDauner thanks for your quick reply! The scenario token is '1e77a04e93c75ba5', but in lots of cases you can notice a distortion in the fusion point of the 2 trajectories.
Since PDM-offset takes as input PDM-Closed trajectory together with the centerline, what could be the reason for this distortion?
from tuplan_garage.
I checked the scenario. The distortion on the cutting point is nothing unusual. When PDM-Offset
predicts large offsets to the PDM-Closed
trajectory, the waypoints on the cutting point are far apart. The nuBoard connects the waypoints with a linear spline, making the output look distorted.
Note that the concept of PDM-Hybrid
purely serves to have a high OLS and CLS simultaneously. The underlying core for planning all comes from PDM-Closed
.
I hope that helps. Let me know if you have further questions!
from tuplan_garage.
@DanielDauner thanks for your clarification! I'll close the issue.
from tuplan_garage.
Related Issues (20)
- AssertionError: Class to be of type <class 'pytorch_lightning.callbacks.base.Callback'>, but is <class 'omegaconf.dictconfig.DictConfig'>! HOT 2
- pdm_closed_planner trajectory's states have no dynamic information HOT 3
- How to accelerate simulation process HOT 1
- PDM-Hybrid that combined PDM-Closed with GC-PGP HOT 2
- accelerate simulation by cuda HOT 3
- calculate inference time HOT 3
- time for training HOT 1
- Centerline and States encoding for PDM_Open and PDM_Offset models HOT 4
- Val14 Dataset HOT 1
- Visualization scripts release HOT 1
- L2 computation issue in open-loop settings HOT 1
- Issue eveluating HOT 4
- Train data and val14 data HOT 1
- about visualization HOT 2
- default_simulation planner HOT 4
- Confused about the checkpoint HOT 1
- Ray Workers Failing Due to Memory Pressure (OOM) in sim_plancnn.sh Script HOT 1
- Problems about the histograms and scenarios of nuboard HOT 6
- Closed-loop simulation error HOT 15
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
π Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. πππ
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google β€οΈ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from tuplan_garage.