Reliability dashboard

This dashboard provides transparency into real-world safety and reliability performance across customer operations. Data is updated quarterly with a one-quarter delay to allow investigations to complete and counts to stabilize.

With this update, we continue to see the flight-wide incident rate improve, driven by a combination of software and hardware improvements and an increase in proportion of flights from dock. We observe that the incident rate is about 3x lower for flights from docks when compared to manual controller-based flights.

Reporting Revision:

This update includes a downward revision of the incident rate from what was published in the first release of the Reliability dashboard. Since we first published the dashboard, we've refined our process on capturing and classifying incidents which identified several gaps that led to us classifying more flights as incidents. Below are the reasons for the revision:

Incident classification was made more conservative

The determination of an incident can be subjective. We have refined our definition and interpretation internally to be more conservative and inclusive (i.e. classify more things as incidents). For example, if an operator lets go too early on hand launch, we consider that an incident consistently now. This led to an increase in the number of overall incidents reported historically.

Root cause analysis can lag incident date by months

Some incidents require deep hardware investigation of the damaged hardware. In the worst case, this can take months to fully land on a clear root cause (esp. if hardware is badly damaged). Previously, these “in process” incidents were excluded pending final analysis, this update now includes these incidents. Going forward, we will continue to include "in process" incidents in our overall incident rate, with best-effort categorization, and we will maintain reporting lag to allow time for root cause analysis.

Incident counting process had gaps internal to Skydio

Various classes of incidents that were either self-diagnosed from the customer as a "human factor" or were otherwise routine environmental factors covered by Skydio Care (i.e. bird strikes) were not correctly categorized and therefore left out of the original incident rate. This process has been resolved and incidents are now reflected on the dashboard in the revised incident rate.

Incident Rate Breakdown:

Technical Factors

• Thin obstacle detection failures remain the largest single technical issue representing more than a quarter of technical factor incidents. Many customers operate in highly challenging environments, including power line inspections, infrastructure mapping, and complex urban environments. Skydio products have world-leading obstacle avoidance, but vision-based navigation is not infallible. Data from these incidents has informed software improvements, which will ship in the coming months to reduce incidents from this factor, but operators can reduce the likelihood by exercising caution around thin obstacles.
  
• Many technical issues that caused incidents in this period have already been addressed. We communicate the most common incident drivers via our Notice to Operator publications. This reporting period included propeller hub failures, battery estimation errors, and flight control system failures, all of which have fixes already deployed or nearing deployment.

Human Factors

• Over 40% of these incidents occur when obstacle avoidance is disabled or unavailable. We also see that about a quarter of these issues happen during takeoff or landing, often due to uneven or unstable surfaces or inconsistent hand takeoff procedures which do not affect dock operations. Many of these incidents could be avoided with greater familiarity with our how to fly Skydio X10 at night & how to launch guidance. Skydio is also investing in reducing operator workload, improving configuration clarity, and increasing the robustness of our autonomy in future releases to further reduce the frequency of these types of incidents.

Environmental Factors

• Bird strikes accounted for roughly half of environmental factor incidents and are an operational reality for small UAS. This is one reason we developed the parachute for at-risk operations.
  
• The other half of incidents were due to icing. Small UAS are highly sensitive to even minor ice accumulation, which is why operations in these conditions fall outside our guidelines. See our guide on operating in winter conditions for best practices and limitations. A software-based icing detection alert introduced in the Battin software release is expected to substantially reduce this risk going forward.

Dock Landing Reliability

• Dock + X10 was achieving 99.86% landing reliability across customer operations at the end of this reporting period and continues to make improvements. Most landings outside of the dock are not crashes, but instead landings at safe landing zones or next to the Dock. These are typically driven by communication or sensor interruptions rather than navigation failures.
  
• In our first six months since launch, we have observed dock-based flights show about a 3x reduction in incident rates, making dock-based operations the most reliable way to perform missions. While this distinction does not affect all failure modes, it provides meaningful benefits by reducing exposure to many takeoff and landing-related human and technical factors.