World model

Overview

A world model predicts how a situation changes when an action is applied 2411.14499 §3.1. The survey separates two jobs. The first is an implicit representation of how the world works. The second is predicting future states, so the model can simulate outcomes for decisions or for data generation 2411.14499 §1 2411.14499 §2. The survey says the definition is still contested 2411.14499 §1.

The invariant the variants share is the one-step transition: a model that maps a state and an action to the next state, and can be rolled forward to look further ahead 2411.14499 §2, Categorization. A text-environment variant keeps this form but outputs the next observation as text and adds a binary success signal 2512.18832 §3.1 eq. 2. Some robot variants also predict a done signal, and use it as a reward or as a termination test 2512.03556 eq. 1 2405.18418 eq. 1.

This page draws on fourteen cards. The survey supplies the family taxonomy and the video and embodied variants. The World Models paper supplies the first end-to-end example, a latent model with an evolved controller. The text-environment study supplies the one empirical result on text environments. The two occupancy papers and a multiview video generator supply the driving forecast variants, with author-reported results on nuScenes. Six newer cards add variants in simulation: a hierarchical humanoid controller, a learned reward inside a video model, search over imagined futures for manipulation, a decoupled driving model for changed vehicles, reward-free pretraining for driving, and a goal-conditioned model trained on episodes between subgoals. A multiplayer study adds a variant that keeps one typed world state shared by all camera views. A survey card covers models of other agents' minds. The survey traces a historical line from Ha and Schmidhuber (2018) and the Dreamer series to video generators 2411.14499 §2 2411.14499 §3.1. The World Models paper has a card and is covered below. The Dreamer series has no card that lists this slug, so it is only named here.

Variants

Latent dynamics model for planning

The model is trained on real transitions and then used to choose actions. The survey casts learning as supervised fitting, either with a squared error or a probabilistic loss 2411.14499 eq. 1 2411.14499 eq. 2. Planning uses the model in three ways: model predictive control over sampled action sequences, Monte Carlo tree search over learned values, and trajectory optimisation in latent space 2411.14499 §3.1. Card: Understanding World or Predicting Future? A Comprehensive Survey of World Models.

Latent model with an evolved controller

The World Models agent has three parts. A VAE encodes each frame into a latent z. A recurrent mixture-density model predicts the next z from the action and its hidden state. A single linear controller maps both to an action 1803.10122 §2 1803.10122 eq. 1. Only the controller sees the reward. The two world-model parts are trained without it 1803.10122 §3.1, and the controller is evolved with CMA-ES, which needs one final score per rollout 1803.10122 §2.3 1803.10122 §A.4. The Car Racing controller has 867 weights 1803.10122 tab. 1. The authors report 906 ± 21 over 100 trials, against 838 ± 11 for a Gym leaderboard entry 1803.10122 tab. 2. Card: World Models.

The same model can also serve as a dream. For VizDoom Take Cover, the controller is trained inside the recurrent model, which also predicts whether the agent dies, and the policy is then run on the real game 1803.10122 §4.1 1803.10122 §4.4. The dream temperature τ changes the outcome. At τ = 0.10 the virtual score is 2,086 ± 140, but the real score is 193 ± 58. The authors say the controller exploits the model at low τ. At τ = 1.15 the real score is 1,092 ± 556 1803.10122 tab. 4 1803.10122 §4.5. The paper does not define precisely how τ acts 1803.10122 §2.2 1803.10122 §4.3. The authors name two limits of the approach: the VAE keeps detail the task does not need, and the recurrent memory has limited capacity 1803.10122 §7.

Hierarchical latent model for humanoid control

Two TD-MPC2 world models are trained separately 2405.18418 §3. A low-level tracker is trained once on noisy rollouts of expert policies for 836 motion-capture clips, then frozen 2405.18418 §3.1 2405.18418 §4.1. A visual high-level planner takes the proprioceptive state and a 64x64 camera image, and outputs a 15-value command for the frozen tracker. It is trained from scratch for each task on the task reward 2405.18418 §3.2 2405.18418 §4.1. On eight simulated humanoid tasks the authors report return comparable to TD-MPC2, except on stairs, where TD-MPC2 scores higher 2405.18418 §4.2 2405.18418 fig. 5. In a 51-person study, 97.8% of the 510 judgements preferred the hierarchical agent's motion 2405.18418 tab. 6. The tracker takes about 12 days on one RTX 3090 2405.18418 §4.1. Card: Hierarchical World Models as Visual Whole-Body Humanoid Controllers.

Model as a reinforcement-learning environment with a learned reward

Here the world model is the training environment for a policy, and it also supplies the reward. The action model takes the policy's 7-DoF end-effector action and returns the next frame and a done signal. A text model writes a goal frame from the instruction, at the step where its own done signal first passes a threshold 2512.03556 §3.1 2512.03556 eq. 6. The reward adds a sparse term, one when the action model's done signal passes a threshold, and a dense term, the LPIPS distance between the current frame and the goal frame 2512.03556 eq. 5 2512.03556 eq. 7 2512.03556 eq. 8. The policy is trained with PPO inside the model 2512.03556 §4.3. With an OpenVLA policy, the authors report 84.1% on out-of-domain pick and place, against 54.5% with the simulator's reward 2512.03556 tab. 1. In-domain differences are small 2512.03556 tab. 1. World-model training took about 96 hours on an H20 cluster 2512.03556 §4.2. All evaluation is in ManiSkill, with no real robot 2512.03556 §5. Card: RoboScape-R: Unified Reward-Observation World Models for Generalizable Robotics Training via RL.

Search over imagined futures at decision time

The model is used inside a search loop, not a fixed rollout. A diffusion-based vision-language-action policy proposes candidate actions at each expanded node. A video model then imagines the outcome of one candidate, and Monte Carlo tree search with a PUCT rule selects the root action with the most visits 2512.18477 §3.3 2512.18477 eq. 5. The policy is CogACT used without further training. The video model is iVideoGPT-medium, fine-tuned on Bridge with a reward head 2512.18477 §4.1. Each decision uses 8 candidates and 8 simulations 2512.18477 §3.1 2512.18477 §4.1. On four SimplerEnv WidowX tasks the average success is 51.0%, against 47.9% for the base policy 2512.18477 tab. 1. The authors call the margin modest 2512.18477 §4.3, and the paper reports no seeds or variance. Card: STORM: Search-Guided Generative World Models for Robotic Manipulation.

Decoupled ego-dynamics model for changed vehicles

This variant splits the model. One part is a parameterised vehicle dynamics model, and the other is an environment model that sees only the ego vehicle's state 2512.02417 §4.1 2512.02417 eq. 9. A learned policy outputs a goal, and a predefined PID controller turns that goal into steering and throttle, so the car's own response is handled by feedback 2512.02417 §4.2.1. Two steps target changed mass or steering limits. At deployment, the original policy's action is replaced by the action that brings a new car to the state the old car would reach 2512.02417 §4.3 2512.02417 eq. 20. During training, imagined rollouts are run under both nominal and shifted parameters 2512.02417 §4.4. On MetaDrive roundabouts the authors report 0.70 success, against 0.53 for DreamerV3. On intersections the paper's model is slightly lower, at 0.88 against 0.91 2512.02417 tab. 1. The mass and steering robustness study is shown only as a figure 2512.02417 fig. 9. Card: Vehicle Dynamics Embedded World Models for Autonomous Driving.

Reward-free pretraining for driving

A DreamerV3-style world model and an explorer are pretrained in CARLA with no task reward. The explorer is rewarded by the disagreement between an ensemble of next-state predictors 2512.18850 §3.2.1 2512.18850 §3.4. The frozen explorer is then run zero-shot on an unseen town. Success averages 73.20% on Town02, against 48.61% for ICM and 55.83% for RND 2512.18850 tab. 6. After 10k fine-tuning steps on task reward, collision avoidance reaches 94.78% on Town02 2512.18850 tab. 6. The results are means over five seeds, with no spread shown 2512.18850 §5. The scope is one simulator, one pair of towns, discrete controls and a semantic-segmentation input 2512.18850 §6. Card: InDRiVE.

Goal-conditioned model trained on subgoal episodes

The world model is goal-conditioned and is trained on data picked to cover the transitions that matter for reaching goals. The paper's method, MUN, adds a second kind of real episode to a goal-conditioned Dreamer-style agent. The agent visits two subgoals drawn from a set of key states, so the model sees backward and cross-trajectory transitions. The policy is then trained only on imagined rollouts, with reward equal to the negative learned temporal distance 2411.02446 §3.1 2411.02446 alg. 1 2411.02446 eq. 7. The subgoals come from farthest-point sampling over the actions in a batch of episodes. The authors use it to pick states where the action changes most, such as grasping and releasing an object 2411.02446 §3.2 [arxiv:2411.02446 alg. 2]. The one-step prediction error is the lowest of the five methods the paper compares on all six tasks 2411.02446 tab. 6. Over 500 simulated trajectories, the paper's model has the lowest compounding error on five tasks. On Fetch Slide, the variant without subgoal selection is slightly lower 2411.02446 tab. 7. On 3-Block stacking the authors report more than 95% success, against about 60% for three baselines within 2.5M steps 2411.02446 §4.3. Each run takes 36 to 70 GPU hours on an A100 2411.02446 tab. 1. Card: Learning World Models for Unconstrained Goal Navigation.

Language-model world knowledge

Language models can carry knowledge about space, time, geography, cities and human behaviour. The survey reports linear representations of space and time, coarse and sometimes inaccurate urban knowledge, and good results on several Theory of Mind tasks with exceptions such as detecting faux pas 2411.14499 §3.2. A second pattern pairs a language model with an explicit model and a planner or verifier; the survey's examples include PDDL domains written by a language model and solved by an off-the-shelf planner 2411.14499 §3.1. Card: Understanding World or Predicting Future? A Comprehensive Survey of World Models.

Mental world model of other agents

The survey's review of Theory of Mind takes the physical world model and extends its state with the agent's own memory and with the observed actions of other agents 2601.02378 §2.1. Two reasoning paradigms are compared. Prompting a language model is easy to apply but works as a black box. Explicit model-based inference is interpretable but costly 2601.02378 §4 2601.02378 tab. 5. The numbers quoted for these methods come from the cited works and were not re-run. One example is 76.6% accuracy for LIMP on MuMA-ToM, against 50.6% for GPT-4o 2601.02378 §4.2. The review argues that most benchmarks are text-only and offline 2601.02378 §5. It runs no experiments of its own. Card: Modeling the Mental World for Embodied AI: A Comprehensive Review.

Fine-tuned language model as text-environment simulator

Here the language model is itself the transition model. It is fine-tuned with supervised learning on logged agent-environment dialogues, and given the dialogue so far and the agent's latest action, it writes the next environment message and a success flag 2512.18832 §3.1 2512.18832 §4. The paper covers ALFWorld, SciWorld, TextWorld, WebShop and StableToolBench. A fine-tuned Qwen2.5-7B reaches 99.87% exact match on ALFWorld next states and 98.60% on SciWorld. On TextWorld and WebShop it reaches 70.60% and 79.05%, and on StableToolBench 48.90% exact match, against prompt-only baselines of 77.04%, 73.08%, 49.12% and 66.09% on the first four 2512.18832 tab. 1. Exact match is strict, and the paper calls its TextWorld figure a lower bound 2512.18832 §4.

Consistency is the harder test. The paper runs an agent in the real environment and in the simulator, then replays the simulated actions in the real environment. The ratio of those success rates is around 0.91 to 0.96 on the three structured environments and 0.67 on WebShop for the Qwen model, averaged over seven API agents 2512.18832 §5.2 2512.18832 tab. 2. The simulator is less reliable for agents it was not trained on. The weakest agent falls below 70% consistency, and mixing data from several agents raises one agent's ratio from 0.49 to 0.81 2512.18832 §5.2 2512.18832 tab. 3.

The paper also reports three agent uses. Pre-execution checks in WebShop reach 36.70% success at 50 checks, against 29.36% with no checks for GPT-4o 2512.18832 tab. 4. Synthetic trajectories are competitive with real ones on SciWorld and WebShop, and a warm-start stage helps final success on ALFWorld and SciWorld. Both gains appear only as plots, with no numbers in the text 2512.18832 §7 2512.18832 fig. 6 2512.18832 fig. 7. Simulated success is the model's own prediction, not a check against the real environment 2512.18832 app. A.2. Card: From Word to World: Can Large Language Models be Implicit Text-based World Models?.

Video generator as simulator

Video generators such as Sora are treated as the main current route to world simulation, with the caveat that it is debated whether they count as world models 2411.14499 §4.1. The desirable capabilities the survey lists are long-term prediction, multimodal input, interactivity and coverage of varied environments. Its named limits are weak causal reasoning and failures on physical laws 2411.14499 §4.1. Card: Understanding World or Predicting Future? A Comprehensive Survey of World Models.

Embodied and action-conditioned environments

Simulators and generated environments for robots and agents are grouped as indoor, outdoor, and dynamic action-conditioned environments. The survey notes that most work so far builds static scenes, and that the trend is toward dynamic, first-person, action-conditioned worlds that respect physical constraints 2411.14499 §4.2. Card: Understanding World or Predicting Future? A Comprehensive Survey of World Models.

Occupancy world model for driving

A camera-only model turns past images into a bird's-eye-view (BEV) state, forecasts how that state changes, and reads occupancy and a trajectory from it. Drive-OccWorld uses a BEVFormer-style encoder and an autoregressive decoder that predicts 3D occupancy and flow over four steps of 0.5 s, which is two seconds on nuScenes 2408.14197 app. C.1 2408.14197 app. C.3. Its memory is normalised by ego motion 2408.14197 eq. 5. A planner samples candidate trajectories and keeps the one with the lowest occupancy cost 2408.14197 §3.4. With the predicted trajectory as the action, the authors report an average open-loop L2 of 0.85 m, against 1.03 m for UniAD under the NoAvg protocol 2408.14197 tab. 5. Card: Driving in the Occupancy World: Vision-Centric 4D Occupancy Forecasting and Planning via World Models for Autonomous Driving.

IR-WM changes what the decoder predicts. It outputs the change to the BEV state, which is added to the previous state 2510.16729 §3.2.2 eq. 6. An alignment step then modulates the features with predicted semantics and ego motion, to limit error build-up over time 2510.16729 §3.2.2. The authors report that alignment adds 0.2 to 0.4 points of per-class IoU on the classes they list 2510.16729 tab. 8, and an average L2 of 0.53 m for planning 2510.16729 tab. 4. The forecast can reach the planner in three ways: tightly coupled, semi-coupled (the default) and fully decoupled 2510.16729 §3.2.4. Card: Vision-Centric 4D Occupancy Forecasting and Planning via Implicit Residual World Models.

The two occupancy papers use nuScenes with occupancy labels 2510.16729 §4.1.1 and train on 8 A100 GPUs 2408.14197 app. C.3 2510.16729 §4.1.3. Their cards record gaps between the paper and the released code. Drive-OccWorld's released configs cannot switch the planner on, and IR-WM's default training path appears to fail as read. Check the cards' Paper vs code sections before you reproduce a planning number.

Multiview video diffusion model with image-based planning rewards

Drive-WM generates the six surround cameras of nuScenes as one video. It uses a latent video diffusion model, conditioned on layout tokens, a text sentence and ego-motion tokens 2311.17918 §3.3 2311.17918 eq. 5. It starts from a single-view image model, adds temporal layers and then multiview layers, and freezes the spatial weights while it trains the rest 2311.17918 §3.1. To keep overlapping views consistent, it generates three reference views first and then three stitched views, each conditioned on its neighbours 2311.17918 §3.2. For planning, a pre-trained planner (VAD) proposes candidates. The model renders each candidate's future, and a reward multiplies a map term and an object term, both measured by perception models run on the generated images 2311.17918 §4.1 2311.17918 §4.2. The reward weights and thresholds are not stated 2311.17918 §4.2.

The authors report FID 15.8 and FVD 122.7 for multiview video, against 16.4 FID for their factorized variant, which raises the matched-point ratio (KPM) from 45.8% to 94.4% 2311.17918 tab. 1 2311.17918 tab. 2. In open-loop planning, the average L2 is 0.80 m and the average collision rate 0.26%, against 1.02 m and 0.93% for VAD with a random command 2311.17918 tab. 3. Most of the gain comes from the object term alone, which gives the same average L2 at a slightly higher collision rate 2311.17918 tab. 4. Training used an A40 with 48 GB and 40k iterations at batch 32 2311.17918 app. B.1. The paper reports no runtime per candidate. Card: Driving into the Future: Multiview Visual Forecasting and Planning with World Model for Autonomous Driving.

Typed shared state for multiplayer worlds

Several players share one world, and each camera is drawn from one state. A learned Logic Engine advances that state by one tick from all players' actions. It is a decoder-only Transformer that predicts each typed record under a schema mask 2608.06257 §3.2 2608.06257 eq. 5. A separate residual U-Net, the Rendering Engine, draws each requested camera from the same predicted state 2608.06257 §3.3 2608.06257 eq. 7. Because every view decodes from that one state, cross-view disagreement is zero by construction, and the authors say the metric therefore says little about the renderer 2608.06257 §4.5. In the 1,024-snake Snake setting the state has 5,121 records per tick 2608.06257 §3.1.

On a matched two-player, two-view Snake benchmark, the authors report parser state recovery of 0.764. The best video baseline, B-PV, scores 0.128, and MultiWorld scores 0.067 2608.06257 tab. 2. The same table gives an invalid rate of 0.177, against 0.052 for the two dense baselines 2608.06257 tab. 2. Full-state exact match on Snake is 41.7% at one tick and 0.0% at 32 and 128 ticks 2608.06257 app. B.2. A typed step takes 45.55 ms, against 3.07 to 3.94 ms for the dense baselines 2608.06257 tab. B.4. Rendering is reported on eight games from held-out reference states, with PSNR from 23.72 to 40.24 dB 2608.06257 tab. 3. The paper's statement of 10,000 recurrent ticks for 1,024 entities has no table in the appendix that the card could find. Its stress runs go up to 2, 4 and 8 snakes 2608.06257 app. C.3. The MultiWorld baseline has 5.60B trainable parameters and MASS's predictor has 5.66M, so the comparison is not parameter-matched 2608.06257 app. B.1. Card: MASS: Multiplayer World Models with Authoritative Shared State.

When to use which

The survey gives no measured comparison between its families, and it states no data or compute requirements for them. The text-environment study reports numbers for one model family on four text environments. The driving papers report numbers on one dataset, and their latency figures use different devices. The multiplayer study's headline comparison uses one game with two players. The simulation papers use different simulators (ManiSkill, CARLA, MetaDrive, SimplerEnv, a humanoid simulator and the six goal-conditioned tasks of the MUN paper), different metrics and different baselines, so their gains cannot be ranked against each other. The table reflects only what each source says.

The World Models pipeline starts from 10,000 random rollouts for its world-model training 1803.10122 §3.2. Data cost matters for the text variant. The training trajectories in the study come from GPT-4o rollouts, at 40K per environment and 70K for WebShop 2512.18832 §4. Data scaling saturates near 20K trajectories on the structured environments, while WebShop keeps improving to about 70K 2512.18832 §6.1 2512.18832 fig. 2. The study's released repository has no license file, so check its terms before reusing its code.

Implementation sketch

The survey gives no algorithm. The loop below combines its model-learning objective and its planning objective. It is our reading of those two parts, not code from the paper 2411.14499 §3.1.

# Fit a one-step transition model to real data  [arxiv:2411.14499 eq. 1]
def fit_world_model(transitions, model):
    for state, action, next_state in transitions:
        predicted = model(state, action)
        update(model, squared_error(next_state, predicted))
    return model

# Score sampled action sequences by predicted return  [arxiv:2411.14499 eq. 3]
def plan(model, reward, state, horizon, samples):
    best_sequence, best_return = None, float("-inf")
    for sequence in sample_sequences(samples, horizon):
        predicted_return = rollout_return(model, reward, state, sequence)
        if predicted_return > best_return:
            best_sequence, best_return = sequence, predicted_return
    # Execution and re-planning are not stated in the survey
    return best_sequence

For the text-environment variant, model is a fine-tuned language model. It takes the dialogue history and the action, and returns the next observation and a success flag 2512.18832 §3.1 2512.18832 eq. 2. The paper's rollout keeps the agent's own history alongside the simulator's history. The paper states no step limit for these rollouts 2512.18832 §3.1.

For the driving variants, the same loop runs over a BEV state rather than a text or symbolic state. IR-WM adds each predicted change to the previous state 2510.16729 eq. 6. Drive-OccWorld scores sampled trajectories against the predicted occupancy at each step, then feeds the chosen trajectory back into the next forecast 2408.14197 §3.4. Drive-WM runs the same loop over generated camera images, and scores each candidate with perception outputs 2311.17918 §4.1 2311.17918 §4.2.

The robot variants that search or train inside the model follow the same shape. STORM expands a tree of policy candidates and gives one unvisited child a model rollout per simulation, then picks the most-visited root action 2512.18477 alg. 1. RoboScape-R uses the model as the environment: each step returns a frame and a done signal, and the reward is computed from them 2512.03556 alg. 1. MASS runs the same loop over a list of typed records rather than a flat state. It decodes each record under a schema mask, assembles the next state, and renders every camera from that one state 2608.06257 alg. 1.

The World Models controller does not plan with the model. It is scored by rollouts in the real environment or in the dream, and CMA-ES updates its weights from those scores 1803.10122 §3.2 1803.10122 §A.4. Use the loop above for planners, and this search for a fixed small policy.

Contradictions

• Physical realism of video generators. The introduction says Sora adheres to physical laws, and a related claim says it does so "perfectly" 2411.14499 §1. Later sections report weak causal reasoning and failures on physical laws 2411.14499 §4.1, and the open-problems section lists the same gap 2411.14499 §6. The intro's statements are the authors' descriptions, not measured results.
• Benchmark size. A table lists UrbanVideo-Bench as 5.2k urban egocentric clips 2411.14499 tab. 8. The open-problems text says 5.2k drone clips 2411.14499 §6. Check the original benchmark before citing either.
• Consistency of the strongest agents. The text says GPT-4.1, GPT-5 and Claude reliably exceed 90% consistency 2512.18832 §5.2. The table gives GPT-5 and Claude 0.89 and 0.88 on SciWorld, and 0.58 to 0.82 on WebShop across the three 2512.18832 tab. 2.
• Text F1 on StableToolBench. The fidelity paragraph reports 49% F1 2512.18832 §5.1. The table's exact-match column holds 48.90 and 49.25, and its F1 column holds 79.15 and 78.97 2512.18832 tab. 1. Use the table.
• Planning protocols are not comparable. Drive-OccWorld reports an average L2 of 0.85 m under NoAvg and 0.47 m under TemAvg 2408.14197 tab. 5. IR-WM reports 0.53 m next to the NoAvg baseline rows, and does not name its own protocol 2510.16729 tab. 4. Drive-WM averages L2 over 1, 2 and 3 s against a VAD baseline on nuScenes 2311.17918 tab. 3. Do not set these figures side by side.
• Baseline provenance. IR-WM lists Drive-OccWorld numbers in its tables. The paper does not say whether they were re-run or copied 2510.16729 tab. 1. Treat the baseline gaps as the authors' claims until they are checked.
• Driving action signal. Drive-WM's main text defines the action as the ego displacement per step 2311.17918 §3.3 2311.17918 §4.1. Its appendix says the action-based data uses speed and steering angle 2311.17918 app. C.1. The paper does not say which one its model used.
• Dreamer and TD-MPC2 baselines across papers. The hierarchical humanoid paper, the driving papers and the reward-free driving paper each report Dreamer-style or TD-MPC2-style baselines, on different tasks and simulators 2405.18418 §4.1 2512.02417 tab. 1 2512.18850 tab. 6. The numbers do not conflict, but they are not comparable.

Reproduction caveats

The new cards record the same kind of gap between paper and code as the occupancy cards. The humanoid paper's text suggests the high-level command is an absolute end-effector position, while its code adds a scaled offset to the current positions 2405.18418 §1 2405.18418 §3. The reward-free driving paper's equation takes the variance of the ensemble, while its code takes the standard deviation before averaging 2512.18850 §3.2.1 2512.18850 eq. 4. RoboScape-R adds a distance term with no stated sign convention 2512.03556 eq. 5 2512.03556 alg. 1, so a literal implementation can reward moving away from the goal. Check each card's Paper vs code section before you reproduce a number from these papers. The Drive-WM card finds no model, planner or reward code in the linked repository, so its planning numbers cannot be checked against code (Drive-WM card).

The MUN card records the same kind of gap. The released configs for three of the six tasks sample subgoals at random, not with the farthest-point sampling the paper describes 2411.02446 §3.2. The code also decides when a subgoal is reached from a learned distance, so the success signal is an estimate 2411.02446 alg. 1. The paper's training steps and Walker subgoal count differ from its own tables 2411.02446 tab. 1 2411.02446 tab. 2. The repository has no LICENSE file, so check its terms before you reuse its code.

The MASS card finds no official code. Its project page has no code link, and the lab repository found there is a static site. The renderer loss weights and the checkpoint-selection weights are not stated, so the typed results cannot be checked against source 2608.06257 app. A.2 2608.06257 app. B.1. (MASS card).

Papers

• World Models: a VAE, a recurrent mixture-density model and a linear controller evolved with CMA-ES. It reports Car Racing scores and a Take Cover policy trained inside the learned model.
• Driving into the Future: Multiview Visual Forecasting and Planning with World Model for Autonomous Driving: a latent video diffusion model that generates six nuScenes camera views from layout, text and ego-motion conditions, and ranks VAD-planned candidates with image-based map and object rewards.
• Hierarchical World Models as Visual Whole-Body Humanoid Controllers: two TD-MPC2 world models, a frozen tracker for 836 motion clips and a visual planner that sets its commands, evaluated on eight simulated humanoid tasks.
• Learning World Models for Unconstrained Goal Navigation: a goal-conditioned Dreamer-style model trained on episodes between subgoals chosen by farthest-point sampling over actions, evaluated on six simulated goal-conditioned tasks.
• Understanding World or Predicting Future? A Comprehensive Survey of World Models: a survey that frames world models as implicit representation and future prediction, covering model-based RL, language-model knowledge, video generators, embodied environments and applications. It reports no experiments of its own.
• From Word to World: Can Large Language Models be Implicit Text-based World Models?: fine-tunes an LLM to predict next observations in text environments, and tests its fidelity, consistency and use for agent training and checks.
• Driving in the Occupancy World: Vision-Centric 4D Occupancy Forecasting and Planning via World Models for Autonomous Driving: a camera-only driving world model that forecasts 4D occupancy and flow, with action conditioning and a cost-based planner.
• Vision-Centric 4D Occupancy Forecasting and Planning via Implicit Residual World Models: predicts the residual change of a BEV state instead of the full future scene, adds feature alignment, and plans from the forecast.
• Vehicle Dynamics Embedded World Models for Autonomous Driving: splits a Dreamer-style model into ego-vehicle and environment parts, with a learned goal policy, a PID controller, and adjustment and augmentation for changed vehicle parameters.
• RoboScape-R: Unified Reward-Observation World Models for Generalizable Robotics Training via RL: trains robot policies with PPO inside an action-conditioned video model, using a reward from a goal frame and a done signal.
• STORM: Search-Guided Generative World Models for Robotic Manipulation: wraps a diffusion VLA policy in Monte Carlo tree search, with a fine-tuned video model scoring imagined outcomes.
• InDRiVE: reward-free driving pretraining with ensemble disagreement as the intrinsic reward, tested zero-shot and few-shot in CARLA towns.
• Modeling the Mental World for Embodied AI: A Comprehensive Review: a survey separating physical from mental world models, comparing Theory of Mind reasoning paradigms and cataloguing benchmarks. It runs no experiments.
• MASS: Multiplayer World Models with Authoritative Shared State: a learned Logic Engine advances one typed world state for all players, and a separate Rendering Engine draws each camera from it, evaluated on a two-player Snake benchmark and on eight games for rendering.