We would like to thank all of the reviewers for their time and thoughtful comments on our paper. MAAC's use of critic-attention only to reduce state-space representation, not To substantiate this claim, we performed an analysis of TarMAC's More importantly, SARNet's use of a dedicated memory unit and the ability's suggestion we will add results from We have described it in Appendix A.1.4 and we will add further details by including However, we did not see performance gains for the tasks in the paper. We will note results with gates in the revision. However, SARNet's performance is substantially better than baselines when the task becomes harder (more agents) and

The golden age for artificial intelligence may have just dawned, but the course is not without its challenges. The AI technology is being tested in the wild before it's been properly vetted in the lab. And, in other cases, even carefully crafted AI systems tend to act in ways that their developers never anticipated. Current AI technology is still far from being able to re-design itself in any significant sense. Even now, though, things with AI may go wrong.

We explore a collaborative multi-agent reinforcement learning setting where a team of agents attempts to solve cooperative tasks in partially-observable environments. In this scenario, learning an effective communication protocol is key. We propose a communication architecture that allows for targeted communication, where agents learn both what messages to send and who to send them to, solely from downstream task-specific reward without any communication supervision. Additionally, we introduce a multi-stage communication approach where the agents co-ordinate via multiple rounds of communication before taking actions in the environment. We evaluate our approach on a diverse set of cooperative multi-agent tasks, of varying difficulties, with varying number of agents, in a variety of environments ranging from 2D grid layouts of shapes and simulated traffic junctions to complex 3D indoor environments. We demonstrate the benefits of targeted as well as multi-stage communication. Moreover, we show that the targeted communication strategies learned by agents are both interpretable and intuitive.

A group of Turkish engineers adapted an ordinary BMW into a Transformer that turns into a giant robot with arms and legs at the touch of a button. In a process lasting seconds, the car doors open and extend into two blade-like arms. Finally, a head emerges from the car roof, creating a fiercesome-looking machine that towers over the tarmac. The ANTIMON looks like any ordinary BMW. It took a team of 12 engineers and four supporting technicians from Turkish firm Letvision built the fully-working prototype, called ANTIMON.

He had just flown into Augusta, Georgia for the Masters. Bad weather caused a pileup of arrivals. The airport ran out of open gates. His flight was stuck on the tarmac. If air traffic controllers had access to Watson, they might have been able to forecast the flight congestion.