Marta Vallejo

Genetic algorithm evaluation of green search allocation policies in multilevel complex urban scenarios

Abstract This paper investigates the relationship between the underlying complexity of urban agent-based models and the performance of optimisation algorithms. In particular, we address the problem of optimal green space allocation within a densely populated urban area. We find that a simple monocentric urban growth model may not contain enough complexity to be able to take complete advantage of advanced optimisation techniques such as genetic algorithms (GA) and that, in fact, simple greedy baselines can find a better policy for these simple models. We then turn to more realistic urban models and show that the performance of GA increases with model complexity and uncertainty level.

Evolving Urbanisation Policies - Using a Statistical Model to Accelerate Optimisation over Agent-based Simulations

Agent-based systems are commonly used in the geographical land use sciences to model processes such as urban growth. In some cases, agents represent civic decision-makers, iteratively making decisions about the sale, purchase and development of patches of land. Based on simple assumptions, such systems are able broadly to model growth scenarios with plausible properties and patterns that can support decision-makers. However, the computational time complexity of simulations limits the use of such systems. Attractive possibilities, such as the optimisation of urban growth policies, tend to be unexplored since the time required to run many thousands of simulations is unacceptable. In this paper we address this situation by exploring an approach that makes use of a statistical model of the agent-based system’s behaviour to inform a rapid approximation of the fitness function. This requires a limited number of prior simulations, and then allows the use of an evolutionary algorithm to optimise urban growth policies, where the quality of a policy is evaluated within a highly uncertain environment. The approach is tested on a typical urban growth simulation, in which the overall goal is to find policies that maximise the ’satisfaction’ of the residents. We find that the model-driven approximation of the simulation is effective at leading the evolutionary algorithm towards policies that yield vastly better satisfaction levels than unoptimised policies.

A fast approximative approach for the Vehicle Routing Problem

In this paper we introduce a three-step heuristic for a complex version of the Vehicle Routing Problem. The Vehicle Routing Problem is focused on the design of optimal delivery routes. A new memory-based approach is developed in order to gather highly valuable experience to predict the best routes in advance. A clustering representation is proposed to allow the system to achieve a simplified abstraction of the model and making use of the pre-existing knowledge to solve more efficiently real instances of the problem. With these elements a new approximative fast approach is developed. The process transforms the VRP in a set of more simple Travelling Salesman Problems which are remarkably easier to solve. A comparison between the results achieved and the application of a genetic algorithm technique is performed. The obtained results demonstrate a noteworthy improvement in terms of performance and time consumption in relation to previous approximative approaches.

Robotic devices and brain-machine interfaces for hand rehabilitation post-stroke

OBJECTIVE To review the state of the art of robotic-aided hand physiotherapy for post-stroke rehabilitation, including the use of brain-machine interfaces. Each patient has a unique clinical history and, in response to personalized treatment needs, research into individualized and at-home treatment options has expanded rapidly in recent years. This has resulted in the development of many devices and design strategies for use in stroke rehabilitation. METHODS The development progression of robotic-aided hand physiotherapy devices and brain-machine interface systems is outlined, focussing on those with mechanisms and control strategies designed to improve recovery outcomes of the hand post-stroke. A total of 110 commercial and non-commercial hand and wrist devices, spanning the 2 major core designs: end-effector and exoskeleton are reviewed. RESULTS The growing body of evidence on the efficacy and relevance of incorporating brain-machine interfaces in stroke rehabilitation is summarized. The challenges involved in integrating robotic rehabilitation into the healthcare system are discussed. CONCLUSION This review provides novel insights into the use of robotics in physiotherapy practice, and may help system designers to develop new devices.

SOPHIA: Soft Orthotic Physiotherapy Hand Interactive Aid

This work describes the design, fabrication and initial testing of a Soft Orthotic Physiotherapy Hand Interactive Aid (SOPHIA) for stroke rehabilitation. SOPHIA consists of: 1.) A soft robotic exoskeleton, 2.) A microcontroller based control system driven by a Brain-Machine Interface (BMI), and 3.) A sensorised glove for passive rehabilitation. In contrast to other rehabilitation devices, SOPHIA is the first modular prototype of a rehabilitation system that is capable of three tasks: aiding extension based assistive rehabilitation, monitoring patient exercises, and guiding passive rehabilitation. Our results show that this prototype of the device is capable of helping healthy subjects to open their hand. Finger extension is triggered by a command from the BMI, while using a variety of sensors to ensure a safe motion. All data gathered from the device will be used to guide further improvements to the prototype, aiming at developing specifications for the next generation device, which could be used in future clinical trials.

Agent-based Modelling for Green Space Allocation in Urban Areas

The task of green space allocation in urban areas consists of identifying a suitable site for allocating green areas. In this proposition paper we discuss about a number of factors like crowdedness, design, distribution and size that could discourage inhabitants to visit a certain green urban area. We plan to cluster our urban residents into several population segments using an Agent-Based Model and study the system in different predefined scenarios. The overall objective of this work is to provide spatial guidance to planners, policy makers and other stakeholders, and shed light on potential policy conflicts among standard policy criteria and user preferences. We will evaluate this potential within a targeted stakeholder workshop.

Evolving Optimal Spatial Allocation Policies for Complex and Uncertain Environments

Urban green spaces play a crucial role in the creation of healthy environments in densely populated areas. Agent-based systems are commonly used to model processes such as green-space allocation. In some cases, this systems delegate their spatial assignation to optimisation techniques to find optimal solutions. However, the computational time complexity and the uncertainty linked with long-term plans limit their use. In this paper we explore an approach that makes use of a statistical model which emulates the agent-based system’s behaviour based on a limited number of prior simulations to inform a Genetic Algorithm.

Combining Soft Robotics and Brain-Machine Interfaces for Stroke Rehabilitation

Stroke is a devastating condition with profound implications for health economics and resources worldwide. Recent works showed that the use of brain-machine interfaces (BMI) could help movement improvements in severely affected chronic stroke patients. This work shows the feasibility and use of a Soft Orthotic Physiotherapy Hand Interactive Aid (SOPHIA) system, able to provide more intense rehabilitation sessions and facilitate the supervision of multiple patients by a single Physiotherapist. The SOPHIA device is controlled by a BMI system and has a lightweight design and low cost. Tests with researchers showed that the system presents a reliable and stable control, besides being able to actively open the volunteers’ hands.

A Multi-Objective Approach to Predicting Motor and Cognitive Deficit in Parkinson's Disease Patients

Parkinsons disease (PD) is a chronic neurodegenerative condition. Traditionally categorised as a movement disorder, nowadays it is recognised that PD can also lead to significant cognitive dysfunction including, in many cases, full-blown dementia. Due to the wide range of symptoms, including significant overlap with other neurodegenerative conditions, both diagnosis and prognosis remain challenging. In this paper, we describe our use of a multi-objective evolutionary algorithm to explore trade-offs between polynomial regression models that predict different clinical measures, with the aim of identifying features that are most indicative of motor and cognitive PD variants. Our initial results are promising, showing that polynomial regression models are able to predict clinical measures with good accuracy, and that suitable predictive features can be identified.

Online/offline evolutionary algorithms for dynamic urban green space allocation problems

Urban-planning authorities continually face the problem of optimising the allocation of green space over time in developing urban environments. The problem is essentially a sequential decision-making task involving several interconnected and non-linear uncertainties, and requires time-intensive computation to evaluate the potential consequences of individual decisions. We explore the application of two very distinct frameworks incorporating evolutionary algorithm approaches for this problem: (i) an ‘offline’ approach, in which a candidate solution encodes a complete set of decisions, which is then evaluated by full simulation and (ii) an ‘online’ approach which involves a sequential series of optimisations, each making only a single decision, and starting its simulations from the endpoint of the previous run. We study the outcomes, in each case, in the context of a simulated urban development model, and compare their performance in terms of speed and quality. Our results show that the online version is considerably faster than the offline counterpart, without significant loss in performance.