Resource-Performance Tradeoff Analysis for Mobile Robots

Abstract

The design of mobile autonomous robots is challenging due to the limited on-board resources such as processing power and energy. A promising approach is to generate intelligent schedules that reduce the resource consumption while maintaining best performance, or more interestingly, to trade off reduced resource consumption for a slightly lower but still acceptable level of performance. In this paper, we provide a framework that is automatic and quantitative to aid designers in exploring such resource-performance trade-offs and finding schedules for mobile robots, guided by questions such as "what is the minimum resource budget required to achieve a given level of performance?" The framework is based on a quantitative multi-objective verification technique which, for a collection of possibly conflicting objectives, produces the Pareto front that contains all the achievable optimal trade-offs. The designer then selects a specific Pareto point based on the resource constraints and desired performance level, and a correct-by-construction schedule that meets those constraints is automatically generated. We demonstrate the efficacy of this framework on several robotic scenarios in both simulations and experiments.

Energy-Aware Planning-Scheduling for Autonomous Aerial Robots

“…In view of the availability of powerful heterogeneous computing hardware [19], the use of onboard computations is further expected to increase in the foreseeable future [20]. In this context, planning-scheduling energy awareness is a recent research direction [11], [17], [18], [21]. Early studies (2000-2010) varied hardware-dependent aspects, e.g., frequency and voltage, along with motion aspects, e.g., motor and travel velocities [7], [11], [12], [22] whereas the literature from the past decade derives energy-aware plans-schedules in broader terms.…”

“…Early studies (2000-2010) varied hardware-dependent aspects, e.g., frequency and voltage, along with motion aspects, e.g., motor and travel velocities [7], [11], [12], [22] whereas the literature from the past decade derives energy-aware plans-schedules in broader terms. These include simultaneous considerations for planning-scheduling in perception [17], localization [21], navigation [10], and anytime planning [18]. These studies are focused on ground-based robots [7], [17], [21], [22], yet, aerial robots are particularly affected by energy considerations, as it would be generally required to land to recharge the battery.…”

Reinforcement co-Learning of Deep and Spiking Neural Networks for Energy-Efficient Mapless Navigation with Neuromorphic Hardware

“…Traditionally, navigation has been relying on global knowledge in the form of maps of the environment [1]. Yet, for many applications in need of effective navigation, the construction of an informative map is prohibitively expensive, due to real-time requirements and the limited energy resources [2], [3]. The recent introduction of deep reinforcement learning (DRL) methods, such as deep deterministic policy gradient (DDPG) [4], enabled learning of optimal control policies for mapless navigation, where the agent navigates using its local sensory inputs and limited global knowledge [5], [6], [7].…”

DeepNetQoE: Self-adaptive QoE Optimization Framework of Deep Networks

“…To achieve a balance between the consumption of computing resources and the performance of the model, it is necessary to find an optimal model with a reasonable consumption of resources. For example, in the design of an autonomous mobile robot, Lahijianian et al [10] argue that reducing the consumption of computational resources does not seriously impact the autonomous ability of the robot. This clearly indicates that there is a reasonable tradeoff between resources and performance.…”