ACT: Action Chunking with Transformers

8 min read

Predict coherent sequences of joint targets, then blend overlapping plans into smooth, reactive control for delicate bimanual tasks.

Policy size~80M parametersControl rate50 HzPaper chunk sizek = 100Observation4 RGB cameras + jointsActionk × 14 joint targetsInference~10 ms

the key shift

Predict a short plan, not one twitch

Behavior cloning compounds small errors across a long trajectory. ACT predicts the next k joint targets together, reducing the effective horizon from T decisions to T/k chunks.

Single-step policy

π(aₜ | sₜ)

horizon T
a1a2a3a4a5a6a7a8a9a10

Every prediction becomes another place where error can push the robot away from the demonstration distribution.

Action-chunk policy

π(aₜ:ₜ₊ₖ | sₜ)

horizon T/k
a1a2a3a4a5

A chunk can encode a coherent subskill, such as grasp, lift, and handover, including pauses that a Markovian one-step policy can misread.

Important nuance: the deployed policy is still queried at every control step. Overlapping chunks are combined with temporal ensembling, so the robot remains reactive.

architecture

A CVAE teaches a Transformer to generate action chunks

During training, the future demonstration supplies a latent style variable. The policy combines that style with current images and joint positions, then decodes the whole action sequence in parallel.

1

Infer style

CVAE encoder

[CLS] + current joints qₜ
+ demonstrated actions aₜ:ₜ₊ₖ

μ, σ → z = μ + σ ⊙ ε

A BERT-like encoder compresses the action sequence into a latent style. Images are omitted from this training-only path for speed.

2

Read the scene

Four camera views

ResNet-18 per view, flattened with 2D position embeddings

qₜ + z

State and style

Transformer encoder

Fuses 1,200 visual tokens with projected joint and style tokens into a 1,202 × 512 context sequence.

3

Decode the chunk

Transformer decoder

k fixed positional queries cross-attend to the encoded observation.

âₜ:ₜ₊ₖ ∈ ℝᵏˣ¹⁴

Absolute targets for 6 joints plus one gripper on each arm

cvae

Learn many human styles, deploy one stable style

The latent z explains why equally valid demonstrations can move differently from the same-looking state. It is available from the future only while training.

Training

encoder + policy
Human action chunk
style z
L = ‖a - â‖₁ + β DKL(q(z | a, q) ‖ N(0, I))

Reconstruction preserves the demonstrated chunk. KL regularization makes the style space smooth and keeps it near a unit Gaussian.

Inference

policy only
Images + joints
z = 0
CVAE encoder is discarded

Zero is the mean of the training prior, so the deployed policy chooses a deterministic default behavior instead of sampling a new style.

closed-loop execution

Overlap the plans, then agree on this step

A fresh chunk is predicted at every timestep. Several chunks therefore propose an action for the same physical time, and ACT blends those proposals rather than switching plans abruptly.

policy call
t−2
t−1
t
t+1
t+2
query t−2
query t−1
·
query t
·
·

The highlighted column contains three predictions for action aₜ, all made from different observation times.

aₜ = Σᵢ wᵢ Âₜ[i] / Σᵢ wᵢ

The paper uses exponential weights wᵢ = exp(−m · i). Unlike ordinary smoothing across neighboring timesteps, this averages predictions for the same timestep, avoiding temporal bias.

evidence

The gains come from all three pieces

The paper isolates chunking, the CVAE objective, and temporal ensembling, then evaluates ACT on six contact-rich real-world tasks.

1% → 44%

action chunking

Average simulated success rose from k = 1 to k = 100.

2% → 35.3%

CVAE on human data

Success without versus with the CVAE objective.

+3.3%

temporal ensemble

Best ACT success with overlapping predictions combined.

Final real-world task success

25 evaluations per task in the paper

50 demos, except Thread Velcro: 100
Slide Ziploc88%
Slot Battery96%
Open Cup84%
Thread Velcro20%
Prep Tape64%
Put On Shoe92%

takeaways

The useful intuition

Chunk, but keep looking

Large chunks shorten the learning horizon. Replanning every step and ensembling overlaps preserve visual feedback and smoothness.

Model human variation

The CVAE prevents different valid demonstration styles from collapsing into a hesitant average. At deployment, z = 0 keeps behavior repeatable.

Know the boundary

Original ACT is trained from scratch per task, has no language input, and still struggles when perception or hardware precision is the bottleneck.

In one sentence: ACT learns a coherent distribution over future joint targets, then repeatedly blends overlapping action chunks so a long, delicate task becomes shorter to learn without becoming open-loop at execution time.

The mental model

ACT treats robot control as receding-horizon sequence prediction. Instead of deciding one motor target at a time, it predicts a whole action chunk from four camera views and the robot's current joint positions. This shortens the effective learning horizon and lets a Transformer represent coordinated subskills such as grasping, lifting, and handing an object from one arm to the other.

The CVAE's latent variable absorbs variation in human demonstrations during training. At inference, the encoder is removed and the latent is fixed to zero. The policy still runs at every timestep, and temporal ensembling combines all overlapping predictions for the current action. The result is a useful balance: plan in chunks, observe continuously, and execute smoothly.

Sources

Built by Suveen.

www.suveenellawela.com v.2026.5