From b5f03acea50f736c2e7674a326ac74e4def4d91c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Stefan=20Fau=C3=9Fer?= Date: Sat, 13 Jun 2026 17:36:14 +0200 Subject: [PATCH 1/2] added description for Send Terminal Obs Info of Sync node --- docs/NODE_REFERENCE.md | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/docs/NODE_REFERENCE.md b/docs/NODE_REFERENCE.md index 54b6e47e..d130bd59 100644 --- a/docs/NODE_REFERENCE.md +++ b/docs/NODE_REFERENCE.md @@ -36,9 +36,8 @@ func _physics_process(delta): return ``` -`needs_reset` is also set to `true` manually (along with `done`) when finishing the episode due to e.g. game over -conditions. Calling `ai_controller.reset()` will set it to `false` and reset the `AIController`'s internal step counter to 0. +Note that `needs_reset` can be set to `true` manually (along with `done`) when finishing the episode due to e.g. game over conditions. #### How to disable the time-out reset: @@ -115,6 +114,10 @@ Speeds the physics up in the environments to enable faster training. The path to a trained .onnx model file to use for inference (only needed for the `Onnx Inference` control mode). +#### Send Terminal Obs Info + +When enabled, then the terminal observation is sent to the Python server on termination or truncation. By default, truncation is triggered automatically when a predefined number of steps have been exceeded, [see `Reset After`](#reset-after) of the `AIController` nodes. However, enabling this allows the Python server to correctly calculate the bootstrap estimations on truncated episodes by utilizing the terminal observations. Alternatively, a truncation can be triggered manually by setting both flags `truncated` and `needs_reset` to `true` and then handling `needs_reset`. + ## Sensors ### RayCastSensor From d43bd135a30491666fe5ea3c08ac5cae656ce82a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Stefan=20Fau=C3=9Fer?= Date: Sun, 21 Jun 2026 20:57:28 +0200 Subject: [PATCH 2/2] Update docs/NODE_REFERENCE.md Co-authored-by: Ivan-267 <61947090+Ivan-267@users.noreply.github.com> --- docs/NODE_REFERENCE.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/NODE_REFERENCE.md b/docs/NODE_REFERENCE.md index d130bd59..2697d105 100644 --- a/docs/NODE_REFERENCE.md +++ b/docs/NODE_REFERENCE.md @@ -116,7 +116,7 @@ The path to a trained .onnx model file to use for inference (only needed for the #### Send Terminal Obs Info -When enabled, then the terminal observation is sent to the Python server on termination or truncation. By default, truncation is triggered automatically when a predefined number of steps have been exceeded, [see `Reset After`](#reset-after) of the `AIController` nodes. However, enabling this allows the Python server to correctly calculate the bootstrap estimations on truncated episodes by utilizing the terminal observations. Alternatively, a truncation can be triggered manually by setting both flags `truncated` and `needs_reset` to `true` and then handling `needs_reset`. +When enabled, then the terminal observation is sent to the Python server on termination or truncation. By default, truncation is triggered automatically when a predefined number of steps have been exceeded, [see `Reset After`](#reset-after) of the `AIController` nodes. However, enabling this allows the Python server to correctly calculate the bootstrap estimations on truncated episodes by utilizing the terminal observations. Alternatively, a truncation can be triggered manually by setting `truncated` to true and then handling the env reset (e.g. set `needs_reset` to `true` and then handle `needs_reset`). ## Sensors