Most Powerful Open Source ERP

Aviation Smart Seats: The comfort airlines could get from analyzing in-flight seat data

This article describes possible applications of equipping airplane seats with smart sensors. It suggests Wendelin as possible architecture for so-called "smart seats" along with recommending hardware options. The implementation of a complete smart seat system with common APIs for airlines to plug into their CRM can be achieved in a matter of weeks.
  • Last Update:2016-09-19
  • Version:002
  • Language:en

Smart things are arriving everywhere - including airplane seats. There are numerous application scenarios: health, security, customer satisfaction or maintenance. Implementing a smart seat is trivial: add a few sensors to each seat, leverage existing digital control systems, leverage existing Wifi networks present in airplanes and deploy the Wendelin architecture for data collection and big data. A smart seat could probably be prototyped in mere weeks and deployed as an upgrade in a matter of months.

Application 1: Passenger Health Monitoring

Imagine a long haul flight. Flight attendants and most passengers are sleeping. A passenger who recently underwent surgery suddenly starts shaking and sweating. Immediate attention and first aid may be necessary to prevent this passenger from collapsing and dying on board of an embolism or heart attack. If the crew on duty and nearby passengers react too late or without apprehending the situation, the passenger stands a high risk of dying. Considering the latest flat beds and semi-privatized cabins in business and first class will make detection of an emergency even more unlikely.

Yet, from a sensor standpoint a passenger's movement patterns during a cardiac arrest could be easily identified by either embedding pressure sensors in the seat or indirectly measuring applied pressure through servo-motors. Heat sensors could be useful too.

Aside from emergency situation, these pressure and heat pattern detection methods could be used to detect intoxicated passengers or cases of air rage. With additional incident record management - as already being used on some airlines with on-cabin tablets - pressure patterns could even be used to predict drunkenness before it reaches an uncontrollable or dangerous state.

.Aviation Smart Seats Photo: - kantver

Application 2: Criminal Behavior Prediction

Pressure and heat patterns could also be useful for crime prevention. A passenger performing unusual movements at unusual times could trigger an alarm for the crew to check the situation: Is the passenger just moving in an unusual way, maybe trying to lace his or her shoes in-flight or is he or she taking explosives out of their shoes?

The sensors could also be used to detect multiple passengers standing at the same time and from this predict imminent attack in the cabin. This could be especially useful on long haul flights when most of the crew is on rest.

Application 3: Passenger Satisfaction

Airlines are increasingly conducting onboard passenger surveys. But not all passengers are replying. Suppose passenger seat position, pressure and heat could be correlated to satisfaction or comfort. If this were the case, a smart seat could also function as a real time survey system. A passenger moving and changing position frequently may be a sign of poor passenger satisfaction, whereas a few moves may be a sign of good satisfaction.

By collecting such information through a smart seat, airlines may be able to better adjust the schedule of onboard meals, their timetables or passenger assignment to seats. The seat could even detect whenever the passenger is hungry or thirsty and trigger crew assistance.

Thinking ahead further, airline smart seats could even store passenger preferences. A passenger could then restore his seat to his preferred position by clicking on the "smart position" button of the seat.

Application 4: Predictive Seat Maintenance

Although airplane seat makers believe that their seats are fully reliable, the reality is different. It is quite frequent that a seat motors get blocked or partly blocked. Passengers have to click on the leg angle buttons ten times in order to reach the flat bed position due to malfunctioning of the flatbed defaults. Such small issues and other could easily be prevented by using machine learning applied to the seat position control. Detecting that seat position is too slow or that passenger is trying to click repeatedly on buttons could be a sign a future seat maintenance incident.

Implementing a Prototype Smart Seat

Smart seat implementation is not difficult with Nexedi Enterprise Free Software. The Wendelin IoT management platform provides everything needed on the software side:

  • Reliable data collection (fluentd)
  • Scalable data storage (NEO)
  • Advanced machine learning algorithms (scikit-learn)
  • Big data appstore with standard APIs for airline companies to query data or run A.I. scripts (ERP5)
  • Remote deployment of OS and smart logic into the seat (SlapOS)
  • Automated data center management (SlapOS)
  • Seat-to-seat communication (re6st)

A basic deployment of Wendelin can be finished within a week. A custom deployment with Big Data Appstore can be finished in 3 months. A very large project with data center replication and custom A.I. logic may require up to a year.

The hardware components may be a bit more challenging but the following standards could be feasible:

  • Wifi for communication (seat to big data, seat-to-seat)
  • I2C for sensor connectivity (as long as few sensors are used)

Most modern airplanes already include a Wifi network. Rather than reimplementing yet another network inside the plane, we suggest to reuse it. If this is not acceptable for security reasons, we could then simply consider a seat-to-seat Wifi ad-hoc network (babel/re6st) with a firewalled gateway to the Internet going through the airplane's main internet gateway.

In terms of hardware, the best option could be to upgrade the PCB that already controls servo-motors of the seat. However, for a prototype, it is easier to use prototyping hardware from OlimexSeedstudio or AdafruitJetson TX1 may also be an option to test GPU acceleration of seat A.I. scripts. Setting up prototype is a matter of a few weeks.