While editing new files for Marlin 1.1.0-RCBugFix I came across new settings for auto bed leveling recently added. It appears now to have at least three ways to determine level rather than the previous two. Marlin 1.0.0, upon which all Robo firmwares variants are based (I will not call them versions since they all start with Marlin 1.0.0), can auto-level using 3 arbitrary points on the bed or using a grid. The 3 point auto-leveling is pretty simplistic, it is fast but it depends on the bed being nearly 100% flat in order to work well. Similar in a way in which one would level a transit or camera on a tripod. The grid auto-leveling, which is enabled by default in Robo firmware is a bit more sophisticated and creates a grid of offsets to determine bed leveling, but because of its linear nature it really only levels in one plane. That is perfect if you bed is only tilted but not warped in any way. Hence the reason all of us that have struggled with auto-leveling have insisted that your bed needs to be fairly level in the first place in order for auto-leveling to work properly. The new auto-leveling method introduced sometime in September 2016 is a grid with a bilinear grid. I am not here to discuss the math involved since it is way over my head, suffice it to say that it creates a grid similar to the old grid auto-leveling except it works with beds that are less than flat and may be tilted in multiple directions, essentially a warped bed. Since Robo uses the linear method as well as having the weight of the entire hotend pushing down on the bed before the Z-Axis endstops trigger as well as bed flexing at different rates depending on where you probe, I believe that this is one of the major causes of poor leveling performance. Bottom line, if you want to stay with Robo supplied firmware, make sure you manually level your bed to the absolute best of your ability. If you move to a firmware based on 1.1.0 then make sure that bilinear is the method enabled and make sure you bed is moderately level, or switch to MESH as both are very close to the same thing (simplified explanation).