Herman A. Engelbrecht

F. Bredell, H. A. Engelbrecht, J. C. Schoeman

Traditional multi-agent reinforcement learning (MARL) algorithms, such as independent Q-learning, struggle when presented with partially observable scenarios, and where agents are required to develop delicate action sequences. This is often the result of the reward for a good action only being available after other agents have taken theirs, and these actions are not credited accordingly. Recurrent neural networks have proven to be a viable solution strategy for solving these types of problems, resulting in significant performance increase when compared to other methods. In this paper, we explore a different approach and focus on the experiences used to update the action-value functions of each agent. We introduce the concept of credit-cognisant rewards (CCRs), which allows an agent to perceive the effect its actions had on the environment as well as on its co-agents. We show that by manipulating these experiences and constructing the reward contained within them to include the rewards received by all the agents within the same action sequence, we are able to improve significantly on the performance of independent deep Q-learning as well as deep recurrent Q-learning. We evaluate and test the performance of CCRs when applied to deep reinforcement learning techniques at the hands of a simplified version of the popular card game Hanabi.

F. Bredell, H. A. Engelbrecht, J. C. Schoeman

The card game Hanabi is considered a strong medium for the testing and development of multi-agent reinforcement learning (MARL) algorithms, due to its cooperative nature, partial observability, limited communication and remarkable complexity. Previous research efforts have explored the capabilities of MARL algorithms within Hanabi, focusing largely on advanced architecture design and algorithmic manipulations to achieve state-of-the-art performance for various number of cooperators. However, this often leads to complex solution strategies with high computational cost and requiring large amounts of training data. For humans to solve the Hanabi game effectively, they require the use of conventions, which often allows for a means to implicitly convey ideas or knowledge based on a predefined, and mutually agreed upon, set of "rules" or principles. Multi-agent problems containing partial observability, especially when limited communication is present, can benefit greatly from the use of implicit knowledge sharing. In this paper, we propose a novel approach to augmenting an agent's action space using conventions, which act as a sequence of special cooperative actions that span over and include multiple time steps and multiple agents, requiring agents to actively opt in for it to reach fruition. These conventions are based on existing human conventions, and result in a significant improvement on the performance of existing techniques for self-play and cross-play for various number of cooperators within Hanabi.

Kevin Eloff, Okko Räsänen, Herman A. Engelbrecht et al.

Multi-agent reinforcement learning has been used as an effective means to study emergent communication between agents, yet little focus has been given to continuous acoustic communication. This would be more akin to human language acquisition; human infants acquire language in large part through continuous signalling with their caregivers. We therefore ask: Are we able to observe emergent language between agents with a continuous communication channel? Our goal is to provide a platform to begin bridging the gap between human and agent communication, allowing us to analyse continuous signals, how they emerge, their characteristics, and how they relate to human language acquisition. We propose a messaging environment where a Speaker agent needs to convey a set of attributes to a Listener over a noisy acoustic channel. Using DQN to train our agents, we show that: (1) unlike the discrete case, the acoustic Speaker learns redundancy to improve Listener coherency, (2) the acoustic Speaker develops more compositional communication protocols which implicitly compensates for transmission errors over a noisy channel, and (3) DQN has significant performance gains and increased compositionality when compared to previous methods optimised using REINFORCE.

Benjamin Evans, Herman A. Engelbrecht, Hendrik W. Jordaan

Reinforcement learning (RL) has shown to be a valuable tool in training neural networks for autonomous motion planning. The application of RL to a specific problem is dependent on a reward signal to quantify how good or bad a certain action is. This paper addresses the problem of reward signal design for robotic control in the context of local planning for autonomous racing. We aim to design reward signals that are able to perform well in multiple, competing, continuous metrics. Three different methodologies of position-based, velocity-based, and action-based rewards are considered and evaluated in the context of F1/10th racing. A novel method of rewarding the agent on its state relative to an optimal trajectory is presented. Agents are trained and tested in simulation and the behaviors generated by the reward signals are compared to each other on the basis of average lap time and completion rate. The results indicate that a reward based on the distance and velocity relative to a minimum curvature trajectory produces the fastest lap times.

Benjamin Evans, Hendrik W. Jordaan, Herman A. Engelbrecht

The problem of autonomous racing is to navigate through a race course as quickly as possible while not colliding with any obstacles. We approach the autonomous racing problem with the added constraint of not maintaining an updated obstacle map of the environment. Several current approaches to this problem use end-to-end learning systems where an agent replaces the entire navigation pipeline. This paper presents a hierarchical planning architecture that combines a high level planner and path following system with a reinforcement learning agent that learns that subsystem of obstacle avoidance. The novel "modification planner" uses the path follower to track the global plan and the deep reinforcement learning agent to modify the references generated by the path follower to avoid obstacles. Importantly, our architecture does not require an updated obstacle map and only 10 laser range finders to avoid obstacles. The modification planner is evaluated in the context of F1/10th autonomous racing and compared to a end-to-end learning baseline, the Follow the Gap Method and an optimisation based planner. The results show that the modification planner can achieve faster average times compared to the baseline end-to-end planner and a 94% success rate which is similar to the baseline.

André Nortje, Herman A. Engelbrecht, Herman Kamper

Standard video codecs rely on optical flow to guide inter-frame prediction: pixels from reference frames are moved via motion vectors to predict target video frames. We propose to learn binary motion codes that are encoded based on an input video sequence. These codes are not limited to 2D translations, but can capture complex motion (warping, rotation and occlusion). Our motion codes are learned as part of a single neural network which also learns to compress and decode them. This approach supports parallel video frame decoding instead of the sequential motion estimation and compensation of flow-based methods. We also introduce 3D dynamic bit assignment to adapt to object displacements caused by motion, yielding additional bit savings. By replacing the optical flow-based block-motion algorithms found in an existing video codec with our learned inter-frame prediction model, our approach outperforms the standard H.264 and H.265 video codecs across at low bitrates.

André Nortje, Willie Brink, Herman A. Engelbrecht et al.

Recent deep learning models outperform standard lossy image compression codecs. However, applying these models on a patch-by-patch basis requires that each image patch be encoded and decoded independently. The influence from adjacent patches is therefore lost, leading to block artefacts at low bitrates. We propose the Binary Inpainting Network (BINet), an autoencoder framework which incorporates binary inpainting to reinstate interdependencies between adjacent patches, for improved patch-based compression of still images. When decoding a patch, BINet additionally uses the binarised encodings from surrounding patches to guide its reconstruction. In contrast to sequential inpainting methods where patches are decoded based on previons reconstructions, BINet operates directly on the binary codes of surrounding patches without access to the original or reconstructed image data. Encoding and decoding can therefore be performed in parallel. We demonstrate that BINet improves the compression quality of a competitive deep image codec across a range of compression levels.

Ryan Eloff, Herman A. Engelbrecht, Herman Kamper

Imagine a robot is shown new concepts visually together with spoken tags, e.g. "milk", "eggs", "butter". After seeing one paired audio-visual example per class, it is shown a new set of unseen instances of these objects, and asked to pick the "milk". Without receiving any hard labels, could it learn to match the new continuous speech input to the correct visual instance? Although unimodal one-shot learning has been studied, where one labelled example in a single modality is given per class, this example motivates multimodal one-shot learning. Our main contribution is to formally define this task, and to propose several baseline and advanced models. We use a dataset of paired spoken and visual digits to specifically investigate recent advances in Siamese convolutional neural networks. Our best Siamese model achieves twice the accuracy of a nearest neighbour model using pixel-distance over images and dynamic time warping over speech in 11-way cross-modal matching.