NL1044226B1 - System for providing advice on parking a vehicle, in particular a solar vehicle - Google Patents

Abstract

System for providing advice for parking a solar vehicle, comprising first determining means designed to determine one or more sunny parking spaces that are suitable for charging the solar vehicle because of their solar radiation. The system preferably also comprises second determining means, which are designed for determining one or more routes to one or more sunny parking spaces at a minimum distance between the solar vehicle that has one or more parking spaces and/or with a minimum expected travel time to that one or more parking spaces. The system preferably makes use of Parking Guidance and Information (PGI) systems, sensors in and/or on the solar vehicle GIS and EIS systems, etc. The first and/or second determination means preferably comprise means that are designed to carry out self-learning artificial intelligence algorithms, in particular Reinforcement Learning (RL) algorithms.

Description

i

Short designation: System for giving advice for parking a vehicle, in particular a solar vehicle

DESCRIPTION

The invention relates to a system for giving advice for parking a solar vehicle.

IB Invention is not limited to solar vehicles and can be used to give advice for parking vehicles in general, In the case of general vehicles, advice is for nearest and most economical parking space,

In this application, a solar vehicle means a vehicle that can be powered by using solar energy. To this end, solar panels are attached to the surface (usually the top) of the vehicle.

Photovoltaic (PV) cells convert the energy of the sun directly into electrical energy for driving the vehicle. In order to be able to charge the solar vehicle with maximum solar energy cp, the vehicle should preferably be parked in a sunny parking space, which can be difficult if there is only an underground or covered space. Furthermore, the solar vehicles must optimally use time in charging and time must not be lost in, for example, searching for a parking space or parked in places that are not optimal in contributing to the charging of the solar vehicles,

Finding parking spaces in busy areas is a widely recognized problem. In the modern car-oriented world, many will find it difficult to find

of a parking space when driving to work or in a busy city, For the solar vehicles, this is an extra challenge to find the optimal sunny parking spaces to charge the solar vehicle,

To this end, the invention comprises an automated system which makes it possible to determine the optimum parking location for a particular vehicle in a parking lot or in a city. In this way, the most suitable parking space for charging the solar vehicle is found and, in addition, the search time for a parking space is minimised.

Introduction:

Optimal place and geographical location for charging the solar vehicles; the system advises the driver where there are some parking spaces: — Parking spaces with maximum solar radiation entry, e.g. location gua distance: maybe better to park a little further and walk a bit, but with better charging possibilities {open roof, better sun possibilities);

This information is preferably combined with the following sources in order to provide optimal advice to solar vehicle drivers via a mobile app or integrated information systems in cars: — Information from so-called “Parking Guidance and Information {PGI} syatema™; — Sensors in/on the solar vehicle (light, SPS, LADAR,..}

~ GIS systems, geolocation (geographical location); — Weather information system; ~ Events information system; - Personal usage information such as charging and driving profiles of current user {private database}; - Average usage of comparable bet sonnetig uid private and public databases via ARFI interfaces ; ~ User calendars {optimasl lassdmoment selectable From user's calendar}; 0 — Auto performance information {conder dog information, solar panel charging profile, and solar vehicle usage, solar panel and battery degradation profiles, as well as other critical components associated with can carry san cptimals loading and ri} profile}; 18 — {Parking} payment system, - Any other systems that can contribute to an optimal cij and loading program of the relevant vehicle,

The present solution includes an architect who uses "orowd scurced®" parking related information to calculate the probability of finding optimal parking spaces for the solar vehicular

Depending on specific road segments, parking lots and/or in larger geographic areas, the crowd-sourced parking mapped informations are obtained from veolocdy, in particular vit {(geographically) so-called Vlocationspsren", This approach masks use of a method for mining locale tracks to determine the probability of typically finding parking spaces on specific road segments, parking lots and/or in larger geographic areas.The location tracks can be calculated to classify parking spaces as public, private and semi-

private (for example, only for company employees in certain areas that also include public parking spaces).

The location traces can be recalculated to derive the times and dates {e.g. hours of the day and days of the weekj 3 when a solar vehicle can park at a particular place msg/,

The system will provide real-time traffic availability information for the salt vehicles. The system contains a data

IO base with an inventory of parking spaces in a city, including their geographic location, size and level of demand,

A mobile phone is programmed to access the database to locate a free space (to charge optimally for the solar vehicle), to pay for a

IS requested length of time in the space and to update the database to remove the space from the database of available parking spaces for the devyrgagde length of time,

Frimnary Factors and Invention:

In a world where time becomes an even more precious commodity, searching for an optimal parking space for the solar vehicles can be a major concern and stressor for drivers can be a frustrating task, especially in large cities or large parking areas, Drivers usually have to drive around for a while looking for a space. If they also had to look for an optimal place for charging the solar vehicle {on sun}, that is an extra challenge. System based on the proposed invention can greatly solve this problem, 36

Parking Guidance and information {PGI} systems:

Parking Guidance and Information {PGI} systems or parking guidance and information systems {BBI} are intended to provide motorists with information about the geographic

Physical location, orientation and availability of parking spaces, The BRI systems are expected to redistribute parking demand across the car parks and reduce queues at 3 the most popular car parks,

The following is noted with regard to two systems known from earlier patents, 0S7834778 and US 8884784.

In the patents mentioned above, the systems are limited to giving advice based on actual information such as: geographic location. orientation of parking space and surroundings Zactors such as buildings shadow eto., plus current iS information about occupancy and costs of parking spaces.

The problem with systems using real-time data {as in the cited patents) is that it is almost impossible to give accurate advice on parking availability during arrival time. The data is then outdated and no longer automatic. Advice question by user is usually done he leaves or even sen day earlier. At that time, real-time data about (parking space} is not relevant.

The: self-learning aspect mish in the listed patents. The system according to the present invention is mostly aimed at making predictions based on historical data such as how the solar vehicle and parking spaces are used in combination with self-learning artificial intelligence algorithms.

&

These self-learning artificial intelligence algorithms will give advice on the best parking space for the solar vehicle. Of course, before prediction is given, this advice can be enriched: be {improved} with real-time data (as is the case in the gencemds patent writing tent,

In the system according to the invention, advice is mainly given by interacting with the environment through trial and error and learning with the help of feedback from his own actions and experiences. This learning process also uses historical data such as: - Information from Parking Guidance and Information {PGI} systems {historical data) usage profile of parking space if available; ~ Average usage profiles of comparable parkser places; = Personal use, such as charging and driving profiles of current user {private databases; ~ Average usage profiles of similar zommne vehicle from private and goblin databases via API interfaces; ~ User calendar {optimum loading time to be selected from user's calendar).

In the system according to the present invention use is preferably made of smart self-learning artificial intelligence algorithms such as Reinforcement Learning {EL} algorithms. These are self-learning algorithms that learn from 3} the feedback {feedback} they get from using the system, So every time the system is used, the system improves itself and updates the solar vehicle usage profile and usage profile of the parking space and as a result the system gets better every time (improving performance}. The usage profiles such as those of the solar vehicle and parking space are enriched with average usage profiles from private and 3 public databases if available,

Machine learning artificial intelligence algorithms are algorithms that parse data, learn from dis data, and then apply what they learn to make informed decisions."

Reinforcement Learning {RL} is a type of machine learning technique (self-learning) that allows gene agent to learn through trial and error in an interactive environment with the help of 1§ fesdiack from its own actions and experiences, Reinforcement

Learning (RL) refers to a kind of Machine Leamming methods where the agent receives a delayed reward in the next time tag to evaluate its previous action.

It was mostly used in games (e.g. Atari, Mario}, with performance comparable to or even better than human users, Examples of RL algorithms: Q-learning, Deep Q-learning and SARSA IState-

Action-Reward-State-Action}, Benefits of Reinforcement

Learning {BL)}: ~ Reinforcement Learning {RL} is applicable to a wide variety of complex problems that cannot be addressed by other machine learning algorithms. RL is very close to artificial general intelligence (AGI) because it has the ability to search for a longest-running target while autonomously proving several possibilities at the same time; - Reinforcement Learning (RL} focuses on the problem as a whole. Conventional machine learning algorithms are designed to excel at specific sub-tasks, without any idea of the big picture. RL, on the other hand, does not divide the problem into sub-problems; it works directly to maximize long-term reward Le, It has a clear purpose, understands the goal, and is able to leverage short-term rewards for long-term benefits

Lermijn, — Reinforcement Learning {RL} does not require a separate data collection step, In RL, training”

ID data obtained through the agent's direct interaction with the environment. Training data is the learning agent's experience, not a separate set of data to be fed to the algorithm. This greatly reduces the burden on the supervisor responsible for the training process, = Reinforcement Learning {RL} works in dynamic, uncertain environments, RL algorithms are inherently adaptive and built to respond to changes in the environment. In RL, time matters and the experience the agent: collects is not independent and identically distributed {(i.i.d.), unlike conventional machine learning algorithms. Because the time dimension is deeply buried in the mechanics of RL, learning is inherently adaptive,

Claims (5)

CONCLUSIONS 1. System for providing advice for parking a vehicle, comprising means of determining, based on the geographical location of the vehicle and the geographical location and/or other characteristics of parking spaces, determining and displaying as advice one or more nearby and/or favorable parking spaces in one or more other respects. 2. System for giving advice for parking a vehicle, in particular a solar system, comprising first determining means, which are adapted to determine one or more sunny parking spaces which, due to their solar radiation, are suitable for charging a the vehicle or zaonnevosrbuiyg. 3. System according to claim 1 or 2, additionally comprising second determining means arranged for determining one or more routes to those one or more sun parking spaces at a minimum distance between the vehicle that have one or more parking spaces and and/c with no minimum expected travel time between those & one or more parking spaces. A system according to claim 1, 2 or 3, wherein one or more of said determining means uses one or more of the following means: - one or more Parking Guidance and Information (PGI) systems; — 2#n or more sensors in and/or on the vehicle, such as light, GPS. LADAR sto.: ~ 2 one or more Geographic Information Systems {GIS}; — one or more weasr information systems; — &one or more Events Information Systems {EIS}; - personal and/or vehicle-related usage information such as charging and driving profiles of the user and/or the vehicle; — average use of equal or similar vehicles from private and public databases via API interfaces; — &&n or more of the user's calendars; 12 — vehicle performance information, such as maintenance information, solar panel charging profile, vehicle usage, degradation profiles of solar panels and/or batteries and/or other components that have any bearing on the vehicle's charging and/or driving profile vehicle; — information from one or more parkesr payment systems. A system according to claim 1, 2, 2 or 4, wherein one or more of the determined determining means comprises 283 means adapted to execute self-learning artificial intelligence algorithms, in particular Reinforcement Learning {RL} algorithms,