-
The `Solve` methods are too long, spanning a couple of screens. Should be refactored.
-
# 目的
iLQG(iLQR)でCarを動かす
# 提案内容
矩形領域内に置いた赤黒の障害物を避けたり触ったりするように車を制御する
# タスク
- 車制御用のコスト関数定義
- 収束用にiLQG(iLQR)に正則化を入れる
-
# 目的
iLQGの理解と動作検証の為に、Pendulumを動かしてみる
# 提案内容
一番簡単そうな制御タスクとしてPendulumをiLQGで動かしてみる。
ダイナミクスとコスト関数は、OpenAI GymのPendulumタスクのコードを利用して用意する。
* OpenAI Gym pendulumのstep関数
https://github.com/openai/gym…
-
do you have all versions of gps's implementation?
-
@miyosuda
https://github.com/shigeyukioba/matchernet/blob/de426f6ee87d8057fdf2055f5a177427ad433a22/matchernet/control/mpcenv.py#L74-L77
ここは削除したいです。
前向きシミュレーションの中にシステムノイズを入れたい場合は、
dynamics の…
-
Running `rosrun exotica_examples example_dynamic_time_indexed 10` results in vastly different (read unstable) roll-outs despite using the feedback gains - the first two are essentially the same, the t…
-
If the problem horizon is `T`, then:
- `X` is of size `T`
- `U` is of size `T-1`
- The `solution` in `Solve` in the solvers is **resized** to `T` without initialising or reinitialising to 0. At t…
-
Hello @jgeisler0303,
Thank you very much for sharing your code! I am a MRSD student at CMU's robotic institute. Our team is currently working on an autonomous drift r/c car. We have integrated Yuval …
-
In particular, raw iLQG, CEM, collocation methods.
-
Can someone confirm that these results are normal and expected?
![Negative numbers after +- 8 iterations](https://cloud.githubusercontent.com/assets/3290366/23331797/a5e12f98-fb6d-11e6-9c9e-4ba3824…