Renato Mancuso

Bassel El Mabsout, Shahin Roozkhosh, Siddharth Mysore et al.

Fine-tuning simulation-trained RL agents with real-world data often degrades crucial behaviors due to limited or skewed data distributions. We argue that designer priorities exist not just in reward functions, but also in simulation design choices like task selection and state initialization. When adapting to real-world data, agents can experience catastrophic forgetting in important but underrepresented scenarios. We propose framing live-adaptation as a multi-objective optimization problem, where policy objectives must be satisfied both in simulation and reality. Our approach leverages critics from simulation as "anchors for design intent" (anchor critics). By jointly optimizing policies against both anchor critics and critics trained on real-world experience, our method enables adaptation while preserving prioritized behaviors from simulation. Evaluations demonstrate robust behavior retention in sim-to-sim benchmarks and a sim-to-real scenario with a racing quadrotor, allowing for power consumption reductions of up to 50% without control loss. We also contribute SwaNNFlight, an open-source firmware for enabling live adaptation on similar robotic platforms.

Denis Hoornaert, Cole Strickler, Manos Athanassoulis et al. · harvard

The shift to data-intensive processing from the cloud to the edge has introduced new challenges and expectations for the next generation of intelligent computing systems. As the memory wall continues to grow, modern systems can only meet these performance expectations by displaying data access patterns that exhibit ideal layouts in memory and ideal spatiotemporal locality in caches. However, only a few data-intensive applications are characterized by ideal locality. Instead, most applications exhibit either (i) poor locality when naively implemented and must undergo costly redesigns and tuning or (ii) inflated memory footprint to offer proper locality. To address the aforementioned challenges, we propose a hardware/software co-designed approach that can be implemented on commercially available SoC/FPGA platforms. Our approach seamlessly inserts in the CPUs' data path a Tensor Memory Engine that provides data with an ideal cache locality to running applications by (i) accessing the memory on behalf of the CPUs and (ii) composing a re-organized view of the memory layout. Unlike in- and near-memory computing approaches, it sets itself apart by clearly decoupling computing and memory accesses; computation is still performed on CPUs while the data re-organization is delegated to the Tensor Memory Engine.

Fardin Abdi, Renato Mancuso, Rohan Tabish et al.

Embedded systems in safety-critical environments are continuously required to deliver more performance and functionality, while expected to provide verified safety guarantees. Nonetheless, platform-wide software verification (required for safety) is often expensive. Therefore, design methods that enable utilization of components such as real-time operating systems (RTOS), without requiring their correctness to guarantee safety, is necessary. In this paper, we propose a design approach to deploy safe-by-design embedded systems. To attain this goal, we rely on a small core of verified software to handle faults in applications and RTOS and recover from them while ensuring that timing constraints of safety-critical tasks are always satisfied. Faults are detected by monitoring the application timing and fault-recovery is achieved via full platform restart and software reload, enabled by the short restart time of embedded systems. Schedulability analysis is used to ensure that the timing constraints of critical plant control tasks are always satisfied in spite of faults and consequent restarts. We derive schedulability results for four restart-tolerant task models. We use a simulator to evaluate and compare the performance of the considered scheduling models.

William Koch, Renato Mancuso, Azer Bestavros

Little innovation has been made to low-level attitude flight control used by uncrewed aerial vehicles (UAVs), which still predominantly uses the classical PID controller. In this work we introduce Neuroflight, the first open source neuro-flight controller firmware. We present our toolchain for training a neural network in simulation and compiling it to run on embedded hardware. Challenges faced jumping from simulation to reality are discussed along with our solutions. Our evaluation shows the neural network can execute at over 2.67kHz on an Arm Cortex-M7 processor and flight tests demonstrate a quadcopter running Neuroflight can achieve stable flight and execute aerobatic maneuvers.

William Koch, Renato Mancuso, Richard West et al.

Autopilot systems are typically composed of an "inner loop" providing stability and control, while an "outer loop" is responsible for mission-level objectives, e.g. way-point navigation. Autopilot systems for UAVs are predominately implemented using Proportional, Integral Derivative (PID) control systems, which have demonstrated exceptional performance in stable environments. However more sophisticated control is required to operate in unpredictable, and harsh environments. Intelligent flight control systems is an active area of research addressing limitations of PID control most recently through the use of reinforcement learning (RL) which has had success in other applications such as robotics. However previous work has focused primarily on using RL at the mission-level controller. In this work, we investigate the performance and accuracy of the inner control loop providing attitude control when using intelligent flight control systems trained with the state-of-the-art RL algorithms, Deep Deterministic Gradient Policy (DDGP), Trust Region Policy Optimization (TRPO) and Proximal Policy Optimization (PPO). To investigate these unknowns we first developed an open-source high-fidelity simulation environment to train a flight controller attitude control of a quadrotor through RL. We then use our environment to compare their performance to that of a PID controller to identify if using RL is appropriate in high-precision, time-critical flight control.

Bassel El Mabsout, Abdelrahman Abdelgawad, Renato Mancuso

Practitioners designing reinforcement learning policies face a fundamental challenge: translating intended behavioral objectives into representative reward functions. This challenge stems from behavioral intent requiring simultaneous achievement of multiple competing objectives, typically addressed through labor-intensive linear reward composition that yields brittle results. Consider the ubiquitous robotics scenario where performance maximization directly conflicts with energy conservation. Such competitive dynamics are resistant to simple linear reward combinations. In this paper, we present the concept of objective fulfillment upon which we build Fulfillment Priority Logic (FPL). FPL allows practitioners to define logical formula representing their intentions and priorities within multi-objective reinforcement learning. Our novel Balanced Policy Gradient algorithm leverages FPL specifications to achieve up to 500\% better sample efficiency compared to Soft Actor Critic. Notably, this work constitutes the first implementation of non-linear utility scalarization design, specifically for continuous control problems.

Siddharth Mysore, Bassel Mabsout, Renato Mancuso et al.

Actors and critics in actor-critic reinforcement learning algorithms are functionally separate, yet they often use the same network architectures. This case study explores the performance impact of network sizes when considering actor and critic architectures independently. By relaxing the assumption of architectural symmetry, it is often possible for smaller actors to achieve comparable policy performance to their symmetric counterparts. Our experiments show up to 99% reduction in the number of network weights with an average reduction of 77% over multiple actor-critic algorithms on 9 independent tasks. Given that reducing actor complexity results in a direct reduction of run-time inference cost, we believe configurations of actors and critics are aspects of actor-critic design that deserve to be considered independently, particularly in resource-constrained applications or when deploying multiple actors simultaneously.

Siddharth Mysore, Bassel Mabsout, Kate Saenko et al.

We focus on the problem of reliably training Reinforcement Learning (RL) models (agents) for stable low-level control in embedded systems and test our methods on a high-performance, custom-built quadrotor platform. A common but often under-studied problem in developing RL agents for continuous control is that the control policies developed are not always smooth. This lack of smoothness can be a major problem when learning controllers %intended for deployment on real hardware as it can result in control instability and hardware failure. Issues of noisy control are further accentuated when training RL agents in simulation due to simulators ultimately being imperfect representations of reality - what is known as the reality gap. To combat issues of instability in RL agents, we propose a systematic framework, `REinforcement-based transferable Agents through Learning' (RE+AL), for designing simulated training environments which preserve the quality of trained agents when transferred to real platforms. RE+AL is an evolution of the Neuroflight infrastructure detailed in technical reports prepared by members of our research group. Neuroflight is a state-of-the-art framework for training RL agents for low-level attitude control. RE+AL improves and completes Neuroflight by solving a number of important limitations that hindered the deployment of Neuroflight to real hardware. We benchmark RE+AL on the NF1 racing quadrotor developed as part of Neuroflight. We demonstrate that RE+AL significantly mitigates the previously observed issues of smoothness in RL agents. Additionally, RE+AL is shown to consistently train agents that are flight-capable and with minimal degradation in controller quality upon transfer. RE+AL agents also learn to perform better than a tuned PID controller, with better tracking errors, smoother control and reduced power consumption.

Siddharth Mysore, Bassel Mabsout, Renato Mancuso et al.

A critical problem with the practical utility of controllers trained with deep Reinforcement Learning (RL) is the notable lack of smoothness in the actions learned by the RL policies. This trend often presents itself in the form of control signal oscillation and can result in poor control, high power consumption, and undue system wear. We introduce Conditioning for Action Policy Smoothness (CAPS), an effective yet intuitive regularization on action policies, which offers consistent improvement in the smoothness of the learned state-to-action mappings of neural network controllers, reflected in the elimination of high-frequency components in the control signal. Tested on a real system, improvements in controller smoothness on a quadrotor drone resulted in an almost 80% reduction in power consumption while consistently training flight-worthy controllers. Project website: http://ai.bu.edu/caps