Emerging multi-model workloads with heavy models like recent large language models significantly increased the compute and memory demands on hardware. To address such increasing demands, designing a scalable hardware architecture became a key problem. Among recent solutions, the 2.5D silicon interposer multi-chip module (MCM)-based AI accelerator has been actively explored as a promising scalable solution due to their significant benefits in the low engineering cost and composability. However, previous MCM accelerators are based on homogeneous architectures with fixed dataflow, which encounter major challenges from highly heterogeneous multi-model workloads due to their limited workload adaptivity. Therefore, in this work, we explore the opportunity in the heterogeneous dataflow MCM AI accelerators. We identify the scheduling of multi-model workload on heterogeneous dataflow MCM AI accelerator is an important and challenging problem due to its significance and scale, which reaches O(10^18) scale even for a single model case on 6x6 chiplets. We develop a set of heuristics to navigate the huge scheduling space and codify them into a scheduler with advanced techniques such as inter-chiplet pipelining. Our evaluation on ten multi-model workload scenarios for datacenter multitenancy and AR/VR use-cases has shown the efficacy of our approach, achieving on average 35.3% and 31.4% less energy-delay product (EDP) for the respective applications settings compared to homogeneous baselines.

Sampling-based motion planning (SBMP) algorithms are renowned for their robust global search capabilities. However, the inherent randomness in their sampling mechanisms often result in inconsistent path quality and limited search efficiency. In response to these challenges, this work proposes a novel deep learning-based motion planning framework, named Transformer-Enhanced Motion Planner (TEMP), which synergizes an Environmental Information Semantic Encoder (EISE) with a Motion Planning Transformer (MPT). EISE converts environmental data into semantic environmental information (SEI), providing MPT with an enriched environmental comprehension. MPT leverages an attention mechanism to dynamically recalibrate its focus on SEI, task objectives, and historical planning data, refining the sampling node generation. To demonstrate the capabilities of TEMP, we train our model using a dataset comprised of planning results produced by the RRT*. EISE and MPT are collaboratively trained, enabling EISE to autonomously learn and extract patterns from environmental data, thereby forming semantic representations that MPT could more effectively interpret and utilize for motion planning. Subsequently, we conducted a systematic evaluation of TEMP's efficacy across diverse task dimensions, which demonstrates that TEMP achieves exceptional performance metrics and a heightened degree of generalizability compared to state-of-the-art SBMPs.

The natural interaction between robots and pedestrians in the process of autonomous navigation is crucial for the intelligent development of mobile robots, which requires robots to fully consider social rules and guarantee the psychological comfort of pedestrians. Among the research results in the field of robotic path planning, the learning-based socially adaptive algorithms have performed well in some specific human-robot interaction environments. However, human-robot interaction scenarios are diverse and constantly changing in daily life, and the generalization of robot socially adaptive path planning remains to be further investigated. In order to address this issue, this work proposes a new socially adaptive path planning algorithm by combining the generative adversarial network (GAN) with the Optimal Rapidly-exploring Random Tree (RRT*) navigation algorithm. Firstly, a GAN model with strong generalization performance is proposed to adapt the navigation algorithm to more scenarios. Secondly, a GAN model based Optimal Rapidly-exploring Random Tree navigation algorithm (GAN-RRT*) is proposed to generate paths in human-robot interaction environments. Finally, we propose a socially adaptive path planning framework named GAN-RTIRL, which combines the GAN model with Rapidly-exploring random Trees Inverse Reinforcement Learning (RTIRL) to improve the homotopy rate between planned and demonstration paths. In the GAN-RTIRL framework, the GAN-RRT* path planner can update the GAN model from the demonstration path. In this way, the robot can generate more anthropomorphic paths in human-robot interaction environments and has stronger generalization in more complex environments. Experimental results reveal that our proposed method can effectively improve the anthropomorphic degree of robot motion planning and the homotopy rate between planned and demonstration paths.

This paper presents a novel 3D myopic coverage path planning algorithm for lunar micro-rovers that can explore unknown environments with limited sensing and computational capabilities. The algorithm expands upon traditional non-graph path planning methods to accommodate the complexities of lunar terrain, utilizing global data with local topographic features into motion cost calculations. The algorithm also integrates localization and mapping to update the rover's pose and map the environment. The resulting environment map's accuracy is evaluated and tested in a 3D simulator. Outdoor field tests were conducted to validate the algorithm's efficacy in sim-to-real scenarios. The results showed that the algorithm could achieve high coverage with low energy consumption and computational cost, while incrementally exploring the terrain and avoiding obstacles. This study contributes to the advancement of path planning methodologies for space exploration, paving the way for efficient, scalable and autonomous exploration of lunar environments by small rovers.

We consider the problem of an autonomous agent equipped with multiple sensors, each with different sensing precision and energy costs. The agent's goal is to explore the environment and gather information subject to its resource constraints in unknown, partially observable environments. The challenge lies in reasoning about the effects of sensing and movement while respecting the agent's resource and dynamic constraints. We formulate the problem as a trajectory optimization problem and solve it using a projection-based trajectory optimization approach where the objective is to reduce the variance of the Gaussian process world belief. Our approach outperforms previous approaches in long horizon trajectories by achieving an overall variance reduction of up to 85% and reducing the root-mean square error in the environment belief by 50%. This approach was developed in support of rover path planning for the NASA VIPER Mission.

One key decision for heuristic search algorithms is how to balance exploration and exploitation. In classical planning, novelty search has come out as the most successful approach in this respect. The idea is to favor states that contain previously unseen facts when searching for a plan. This is done by maintaining a record of the tuples of facts observed in previous states. Then the novelty of a state is the size of the smallest previously unseen tuple. The most successful version of novelty search is best-first width search (BFWS), which combines novelty measures with heuristic estimates. An orthogonal approach to balance exploration-exploitation is to use several open-lists. These open-lists are ordered using different heuristic estimates, which diversify the information used in the search. The search algorithm then alternates between these open-lists, trying to exploit these different estimates. This is the approach used by LAMA, a classical planner that, a decade after its release, is still considered state-of-the-art in agile planning. In this paper, we study how to combine LAMA and BFWS. We show that simply adding the strongest open-list used in BFWS to LAMA harms performance. However, we show that combining only parts of each planner leads to a new state-of-the-art agile planner.

Airlift operations require the timely distribution of various cargo, much of which is time sensitive and valuable. However, these operations have to contend with sudden disruptions from weather and malfunctions, requiring immediate rescheduling. The Airlift Challenge competition seeks possible solutions via a simulator that provides a simplified abstraction of the airlift problem. The simulator uses an OpenAI gym interface that allows participants to create an algorithm for planning agent actions. The algorithm is scored using a remote evaluator against scenarios of ever-increasing difficulty. The second iteration of the competition was underway from November 2023 to April 2024. In this paper, we describe the competition and simulation environment. As a step towards applying generalized planning techniques to the problem, we present a temporal PDDL domain for the Pickup and Delivery Problem, a model which lies at the core of the Airlift Challenge.

We would like to enable a collaborative multiagent team to navigate at long length scales and under uncertainty in real-world environments. In practice, planning complexity scales with the number of agents in the team, with the length scale of the environment, and with environmental uncertainty. Enabling tractable planning requires developing abstract models that can represent complex, high-quality plans. However, such models often abstract away information needed to generate directly-executable plans for real-world agents in real-world environments, as planning in such detail, especially in the presence of real-world uncertainty, would be computationally intractable. In this paper, we describe the deployment of a planning system that used a hierarchy of planners to execute collaborative multiagent navigation tasks in real-world, unknown environments. By developing a planning system that was robust to failures at every level of the planning hierarchy, we enabled the team to complete collaborative navigation tasks, even in the presence of imperfect planning abstractions and real-world uncertainty. We deployed our approach on a Clearpath Husky-Jackal team navigating in a structured outdoor environment, and demonstrated that the system enabled the agents to successfully execute collaborative plans.

Holographic Optical Tweezers (HOT) are powerful tools that can manipulate micro and nano-scale objects with high accuracy and precision. They are most commonly used for biological applications, such as cellular studies, and more recently, micro-structure assemblies. Automation has been of significant interest in the HOT field, since human-run experiments are time-consuming and require skilled operator(s). Automated HOTs, however, commonly use point traps, which focus high intensity laser light at specific spots in fluid media to attract and move micro-objects. In this paper, we develop a novel automated system of tweezing multiple micro-objects more efficiently using multiplexed optical traps. Multiplexed traps enable the simultaneous trapping of multiple beads in various alternate multiplexing formations, such as annular rings and line patterns. Our automated system is realized by augmenting the capabilities of a commercially available HOT with real-time bead detection and tracking, and wavefront-based path planning. We demonstrate the usefulness of the system by assembling two different composite micro-structures, comprising 5 $\mu m$ polystyrene beads, using both annular and line shaped traps in obstacle-rich environments.

Consider an agent acting to achieve its temporal goal, but with a "trembling hand". In this case, the agent may mistakenly instruct, with a certain (typically small) probability, actions that are not intended due to faults or imprecision in its action selection mechanism, thereby leading to possible goal failure. We study the trembling-hand problem in the context of reasoning about actions and planning for temporally extended goals expressed in Linear Temporal Logic on finite traces (LTLf), where we want to synthesize a strategy (aka plan) that maximizes the probability of satisfying the LTLf goal in spite of the trembling hand. We consider both deterministic and nondeterministic (adversarial) domains. We propose solution techniques for both cases by relying respectively on Markov Decision Processes and on Markov Decision Processes with Set-valued Transitions with LTLf objectives, where the set-valued probabilistic transitions capture both the nondeterminism from the environment and the possible action instruction errors from the agent. We formally show the correctness of our solution techniques and demonstrate their effectiveness experimentally through a proof-of-concept implementation.