4 présentations

• 
  15h30 - 15h55

  AI-Enhanced Risk Assessment for Safe Human-Robot Collaboration

  • Morteza Jalali Alenjareghi, prés., PhD Student
  • Samira Keivanpour, Polytechnique Montréal
  • Yuvin Adnarain Chinniah, Full Professor

  Human-robot collaboration (HRC) requires risk assessment (RA) to ensure safety. However, conventional RA methods face difficulties such as unrealistic simulations, integration and testing limitations, and lack of dynamic realism and sensors. This study investigates how Artificial Intelligence (AI) can improve RA for HRC. The study evaluates the advantages and drawbacks of AI-based RA methods for HRC. AI-based RA methods can identify hazards and complex system modeling more effectively, but they also have challenges such as inadequate coverage of human-robot interactions, dependence on historical data, low adaptability, design complexity, and neglect of human factors. The study highlights the key aspects of AI-based RA methods, such as adaptability, safety, and accuracy. It also emphasizes the importance of following standards to ensure the validity and consistency of AI-based RA methods for different HRC scenarios. Furthermore, the study provides insights for advancing AI-based RA methods for HRC by reviewing the current state-of-the-art, proposing areas for future research, and recommending comprehensive solutions to address the existing gaps.

• 
  15h55 - 16h20

  Optimisation du Désassemblage Collaboratif Humain-Robot à l'aide des Algorithmes Génétiques: Application au reconditionnement des batteries de véhicules électriques

  • Salma Nabli, prés., ingénierie en génie électrique

  Face aux changements climatiques, l'incitation à utiliser des véhicules électriques par plusieurs pays a mené à une augmentation des batteries usagées, nécessitant un recyclage minutieux pour récupérer leurs composants. La variété des modèles de batteries demande une flexibilité que seule l'intervention humaine peut offrir. Toutefois, ce travail répétitif et difficile présente des risques de sécurité et peut causer des troubles musculosquelettiques chez les opérateurs humains. Afin de surmonter ces problèmes, ce projet vise à maximiser la collaboration entre les capacités humaines et robotiques, en fusionnant l'adaptabilité humaine avec l'efficacité et la précision des robots. L'ordre d'exécution des tâches de désassemblage joue un rôle décisif dans l'amélioration de l'efficacité du processus. L’objectif principal de ce projet consiste à optimiser l’ordre de la séquence de désassemblage, en considérant l'incertitude liée à l'intervention humaine. L’optimisation vise à réduire simultanément le temps total du désassemblage et la consommation énergétique de l’installation industrielle. Pour atteindre cet objectif, la méthodologie adoptée inclut la modélisation du problème, le développement d'un algorithme d'optimisation évolutionnaire et une analyse comparative avec des techniques exactes. Les résultats attendus devraient aboutir à un désassemblage plus optimal et sécuritaire, contribuant aussi à réduire l'empreinte environnementale et facilitant la réutilisation des composants des batteries.

• 
  16h20 - 16h45

  I4Evosim: An Educational Serious Game Integrating Commercial, Planning, and Scheduling Decision-Making in an Engineer-To-Order Competitive Environnement

  • Anas Neumann, prés., Polytechnique Montréal / Université Laval
  • Adnene Hajji, Université Laval
  • Monia Rekik, Université Laval
  • Robert Pellerin, Polytechnique Montréal

  I4Evosim (www.i4evosim.com) is a new educational serious game designed to teach students decision-making in an uncertain Engineer-To-Order (ETO) manufacturing environment. The proposed platform is based on a data model and a two-level decision process addressing both the initial planning and the rescheduling stages of multiple ETO projects. Besides optimizing costs and duration, the model used by I4Evosim includes a proactive scheduling strategy that minimizes the impact of the uncertainty inherent in ETO projects: schedule instability and waste of time and resources due to frequent design evolutions. The platform also integrates a Lamarckian Layered Genetic Algorithm (LLGA) able to quickly find good solutions. I4Evosim encourages students to (i) design a business strategy in a competitive market (in order to gain projects), (i) study BOMs and design documents, (iii) manipulate decisions related to planning and scheduling, and (iv) conduct a retrospective analysis of their performance. Gamification mechanisms make it easy to grasp a complex subject combining numerous and interdependent decisions, constraints, and objectives. The findings of a preliminary experiment conducted in a course on operations and logistics in the era of Industry 4.0, offered by Université Laval, allowed us to measure the platform’s impact on students’ learning process.

• 
  16h45 - 17h10

  Calibration and Validation of Wearable Sensors for Musculoskeletal Disorders: A Review of Standards and Best Practices

  • Morteza Jalali Alenjareghi, prés., PhD Student
  • Firdaous Sekkay, Assistant Professor
  • Samira Keivanpour, Polytechnique Montréal
  • Camélia Dadouchi, Polytechnique Montréal

  The utilization of wearable sensors for monitoring musculoskeletal health among workers has become widespread in both research and industrial settings. Improving human ergonomics is vital, particularly considering the limitations of existing tools, including limited range and resolution or sensitivity to motion and position. The calibration and validation of wearable sensors pose challenges, especially for monitoring musculoskeletal disorders (MSDs), which are crucial for ensuring accurate and reliable data collection. This paper provides a systematic literature review of international standards, as well as best practices related to the calibration and validation processes of wearable sensors. By studying existing standards and methodologies, this review explains key considerations and challenges in achieving optimal sensor performance, including accuracy, precision, and reliability. Furthermore, the paper emphasizes the importance of adhering to standardized protocols for calibration and validation to enhance the validity and comparability of data across different studies and applications. This review offers insights into emerging trends and future directions in the calibration and validation of wearable sensors for monitoring workers to prevent MSDs. The aim of this paper is to facilitate advancements in the development and implementation of wearable sensor technologies for effective MSD management and prevention strategies.