Communication is an essential skill for intelligent agents; it facilitates cooperation and coordination, and enables teamwork and joint problem-solving. However, effective communication is challenging for robots; given a multitude of information that can be relayed in different ways, how should the robot decide what, when, and how to communicate?

As an example, consider the scenario in Fig. 1 where a robot assistant is tasked to provide helpful information to a human driving the blue car in fog (or to explain the robot’s own driving behavior). There are other cars in the scene, which may not be visible to the human. The robot, however, has access to sensor readings that reveal the environment and surrounding cars. To prevent potential collisions, the robot needs to communicate relevant information — either using a heads-up display or verbally — that intuitively, should take into account what the human driver currently believes, what they can perceive, and the actions they may take.

Fig 1. Robot Assistant needs to provide information to help a human driving the blue car in dense fog. The human has limited visibility and the assistant can highlight cars on a heads-up-display or provide verbal cues (as the human is unable to see highlighted cars that are in the rear)

Prior work on planning communication methods in HRI typically rely on human models, which are typically handcrafted using prior knowledge [1] or learned from collected human demonstrations [2]. Unfortunately, handcrafted models do not easily scale to complex real-world environments with high-dimensional observations, and data-driven models typically require a large number of demonstrations to generalize well. In this work, we seek to combine prior knowledge with data in a manner that reduces both manual specification and sample complexity.

In this work, we seek to combine prior knowledge with data in a manner that reduces both manual specification and sample complexity. Our key insight is that learning differences from a suitable reference model is more data-efficient than learning an entire human model from scratch. We take inspiration from social projection theory [3], which suggests that humans have a tendency to expect others to be similar to ourselves, i.e., a person understands other individuals using one’s self as a reference. We leverage latent state-space models obtained via deep reinforcement learning as the initial model (robot’s self model). To capture the difference between robot and human models, we strategically place learnable implants, which are small functions that can be quickly optimized via gradient-based learning. We call our framework Model Implants for Rapid Reflective Other-agent Reasoning/Learning (MIRROR).

Fig 2. MIRROR’s self-model is a multi-modal latent state-space model (MSSM). In the above, circle nodes represent random variables and shaded nodes are observed during learning.

In MIRROR, the robot’s self-model is a multi-modal state-space model (Fig.2) and is optimized via ELBO loss:

Fig 3. An Example Thresholding Filter as a Perceptual Implant. The (generated) range observations (blue bars) are passed through a thresholding filter to eliminate observations that are beyond a parameterized distance (red bars) in each segment from the human.

Given the latent states z sampled from the inference network, we leverage RL to learn optimal policies. Our approach is based on Stochastic Latent Actor Critic [4]. Once the self-model is trained, we augment it with implanted functions h(·) (e.g. perceptual implant and policy implant). Fig.3 gives an example of the perceptual implant. Given an implant hχ parameterized by χ, we can learn χ by minimizing the following loss given data:

Finally, the robot can optimize communication to maximize task rewards while minimizing communication costs via forward simulation and communication pathways (Fig.4).

Fig 4. Human-Robot Communication via Forward Simulation. In brief, the robot’s self-model is used to simulate the environment and the human model is used to simulate actions. Be leveraging the learnt dynamics network f and inference network g, the two models are coupled and used to “imagine” possible futures. The communication pathways (in teal) serves to link the robot’s communication actions (filtered generated observations) to the human model’s observations. The multi-modal models support M potential communication pathways and the robot may choose one or more of these pathways at any given time step.

Experiments in three simulated domain show that MIRROR is able to learn better human models faster compared to behavioral cloning (BC) and a state-of-the-art imitation learning method. In addition, we report on a human-subject study using the CARLA simulator [5], which reveals that MIRROR provides useful assistive information, enabling participants to complete a driving task with fewer collisions in adverse visibility conditions.

Fig 5. Gridworld Driving where the blue vehicle is moving on a road at constant speed and has to avoid the other red vehicles. In the Fog setting, visibility is reduced; the black region indicates areas not visible to the human.

Fig 6. Search-&-Rescue task where the agent (blue box) starts at the door and is tasked to rescue a victim at the green goal and bring them back to the door. The obstacle in red can appear in either the top or the bottom path, and the victim’s position is randomly initialized in one of three potential positions. In the Smoke variant, visibility is reduced to a small region around the human.

Fig 7. A Bomb Defusal game where a teleoperated robot has 15 seconds to disarm the bomb by pressing three buttons (one in each stage). The correct button at each stage depends on six visible “terminals” (which change after each button press), the bomb type (not visible to the human, but detectable by the robot) and game rules. The rules differ slightly between the robot training environment and the test environment. The human, who has access to the updated rules, has to confirm the robot’s selection.

Fig 8. CARLA Experiment Setup. (A) The stretch of highway that participants drove along. (B) Participants drove the simulated car using a steering wheel with accelerator and brake pedals. (C) and (D) show the difference in visibility in clear and foggy weather. Both cars are visible in the clear setting. In the fog setting, the car on the left is visible, but the car in the front can barely be seen. The car is equipped with a semantic LIDAR and a driving assistant that can provide both visual and verbal cues. Specifically, the agent could highlight selected vehicles through visual bounding boxes and/or provide informative speech (as previously shown in Fig. 1.


Code for reproducing experiments presented in this paper can be found at this Github repo.


If you find our code or the ideas presented in our paper useful for your research, consider citing our paper.

Kaiqi Chen, Jeffrey Fong, and Harold Soh. “MIRROR: Differentiable Deep Social Projection for Assistive Human-Robot Communication” Robotics: Science and Systems, 2022.

    title={MIRROR: Differentiable Deep Social Projection for Assistive Human-Robot Communication}, 
    author={Chen, Kaiqi and Fong, Jeffrey and Soh, Harold},
    booktitle = {Proceedings of Robotics: Science and Systems}, 
    year      = {2022}, 
    month     = {June}}


If you have questions or comments, please contact Kaiqi Chen.


This research is supported by the National Research Foundation Singapore under its AI Singapore Programme (Award Number: AISG-RP-2019-011).


[1] A. Tabrez, S. Agrawal, and B. Hayes, “Explanation-based reward coaching to improve human performance via reinforcement learning,” in 2019 14th ACM/IEEE International Conference on Human-Robot Interaction (HRI). IEEE, 2019, pp. 249–257.

[2] S. Reddy, A. D. Dragan, and S. Levine, “Sqil: Imitation learning via reinforcement learning with sparse rewards,” in International Conference on Learning Representations, 2019.

[3] F. H. Allport, Chapter 13: Social Attitudes and Social Consciousness, ser. Social Psychology. Houghton Mifflin Company, 1924.

[4] T. Haarnoja, A. Zhou, P. Abbeel, and S. Levine, “Soft actor-critic: Offpolicy maximum entropy deep reinforcement learning with a stochastic actor,” in International conference on machine learning. PMLR, 2018, pp. 1861–1870.

[5] A. Dosovitskiy, G. Ros, F. Codevilla, A. Lopez, and V. Koltun, “CARLA: An open urban driving simulator,” in Proceedings of the 1st Annual Conference on Robot Learning, 2017, pp. 1–16.

Written by

Kaiqi Chen