Mobile Kiosk Task Orchestration With Low-Energy ML Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mobile kiosk systems in electronic gaming environments lack flexibility, face challenges in task assignment to capable kiosks, and require inefficient energy management, particularly in casino environments with limited network access.
Innovation Solution
A mobile kiosk management system that assigns tasks efficiently to capable kiosks and manages energy use minimally, using pre-trained datasets and external processing for machine learning functions, and operates with a minimalist approach to extend battery life.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If mobile kiosks perform multiple tasks with full machine learning processing locally, then task capability and versatility improve, but energy consumption increases and battery life decreases
Solution Approach 1:
The machine learning processing is segmented into two parts: pre-training done externally and inference done locally on the mobile kiosk. This allows the system to maintain task capability while reducing energy consumption by performing only lightweight inference operations on the device rather than full training cycles.
Solution Approach 2:
Machine learning models are pre-trained externally before being deployed to the mobile kiosk. This preliminary action transfers the computationally intensive training phase to external systems, leaving only the energy-efficient inference phase for execution on the battery-powered device, thereby extending operational duration.
2Adaptability or versatility
If mobile kiosks operate autonomously with full processing capabilities, then task performance flexibility improves, but device complexity increases
Solution Approach 1:
An external system acts as an intermediary for machine learning model training and updates. The mobile kiosk receives pre-trained models from this external source, allowing it to perform flexible tasks without needing to maintain complex training infrastructure locally, thus reducing device complexity while preserving adaptability.
3Adaptability or versatility
If mobile kiosks use extensive local storage for datasets and models, then machine learning operation capability improves, but device size and weight increase
Solution Approach 1:
Large datasets and training models are extracted from the mobile kiosk and stored externally. Only essential inference models and small datasets are retained locally, enabling the device to maintain machine learning operation capability while significantly reducing its physical size and weight for mobile deployment.
Data Source
AI summary
A mobile kiosk management (MKM) system including a MKM server including a processor and a memory device storing instructions is described. The instructions, when executed by the processor, cause the processor to upload a plurality of applications to a mobile kiosk, identify a first task for the mobile kiosk to perform, and identify a first application for the mobile kiosk to execute to perform the first task. The instructions also cause the processor to cause the mobile kiosk to execute the first application to perform the first task, identify a second task for the mobile kiosk to perform at the venue, and identify a second application for the mobile kiosk to execute to perform the second task. The instructions further cause the processor to determine that the mobile kiosk has performed the first task and cause the mobile kiosk to execute the second application to perform the second task.