| title | Introduction |
|---|---|
| textbook | #introduction |
what is "autonomy"?
we see various examples of it...
| perception | how do you "see" the world around you? |
| perception | how do you "see" the world around you? |
| sensing | various ways to perceive the world around you (e.g, camera, LiDar) |
| perception | how do you "see" the world around you? |
| sensing | various ways to perceive the world around you (e.g, camera, LiDar) |
| compute | what do you "do" with the information about the world? |
| perception | how do you "see" the world around you? |
| sensing | various ways to perceive the world around you (e.g, camera, LiDar) |
| compute | what do you "do" with the information about the world? |
| motion | do your computations result in any "physical" changes? |
| perception | how do you "see" the world around you? |
| sensing | various ways to perceive the world around you (e.g, camera, LiDar) |
| compute | what do you "do" with the information about the world? |
| motion | do your computations result in any "physical" changes? |
| actuation | what "actions", if any, do you take for said physical changes? |
| perception | how do you "see" the world around you? |
| sensing | various ways to perceive the world around you (e.g, camera, LiDar) |
| compute | what do you "do" with the information about the world? |
| motion | do your computations result in any "physical" changes? |
| actuation | what "actions", if any, do you take for said physical changes? |
| planning | can you do some "higher order" thinking (i.e., not just your immediate next move) |
| Autonomy is the ability to |  |
| Autonomy is the ability to perform given tasks |
|
| Autonomy is the ability to perform given tasks based on the system's perception |
|
| Autonomy is the ability to perform given tasks based on the system's perception without human intervention |
|
| Autonomy is the ability to perform given tasks based on the system's perception without human intervention |
|
| cyber |
| cyber |  |
| cyber | |
 |
| cyber | |
|
 |
| cyber | |
|
|
| physical |
| cyber | |
|
|
| physical |  |
| cyber | |
|
|
| physical | |
 |
| cyber | |
|
|
| physical | |
|
 |
| cyber | |
|
|
| physical | |
|
|
hence, they fall under the class of systems → cyber-physical systems
...are everywhere!
...are everywhere!
the embedded components → interactions with the real world
consider the following example of two cars...
the second car is approaching the first
sensors → constantly gathering data/sensing
sensors → constantly gathering data/sensing
- periodic sensing
on detection (of other car) → quickly compute what to do
on detection (of other car) → quickly compute what to do
- periodic sensing
- computation
take physical action → say by braking in time
take physical action (actuation) → say by braking in time
- periodic sensing
- computation
- actuation
- periodic sensing
- computation
- actuation
- periodic sensing
- computation
- actuation
- periodic sensing
- computation
- actuation
on detection (of other car) → quickly compute what to do
on detection (of other car) → quickly compute what to do
complete computation/actuation → before a deadline
complete computation/actuation → before a deadline
- we see one sensor (maybe LiDAR)
- we see one sensor (maybe LiDAR)
- reality → multiple sensors
- we see one sensor (maybe LiDAR)
- reality → multiple sensors
- cameras, radars, lidar, etc.
multiple sensors in an autonomous vehicle → need to combine them somehow
multiple sensors in an autonomous vehicle → need to combine them somehow
once we have information from the sensors (fused or otherwise)...
-v-
-v-
-v-
...but there is an operating system (OS) in there
-v-
...but there is an operating system (OS) in there
and it includes real-time mechanisms
-v-
note: ekf is versatile; can be used for sensor fusion, slam, etc.
-v-
-v-
like braking, engine control
if we design such a system...
if we design such a system...
so, from the perspective of the autonomous system,
so, from the perspective of the autonomous system,
- where am I?
so, from the perspective of the autonomous system,
- where am I?
- where do I need to go?
so, from the perspective of the autonomous system,
- where am I?
- where do I need to go?
- how do I get there?
so, from the perspective of the autonomous system,
- where am I?
- where do I need to go?
- how do I get there?
- what obstacles may I face?
so, from the perspective of the autonomous system,
- where am I?
- where do I need to go?
- how do I get there?
- what obstacles may I face?
- how do I avoid them?
let's not forget the most important question of all...
how do we tackle these issues:
- where am I?
- where do I need to go?
- how do I get there?
- what obstacles may I face?
- how do I avoid them?
- where am I?
- where do I need to go?
- how do I get there?
- what obstacles may I face?
- how do I avoid them?
-v-
simultaneous localization and mapping
figure out where we are
-v-
how to move in the right direction at the micro level, i.e., find waypoints
-v-
"you only live once"
-v-
"you only look once"
-v-
"you only look once"
object detection model that uses convolutional neural networks (cnns)
-v-
avoid objects in immediate path
-v-
how to get to destination at the macro level → uses waypoints
to run all of these functions
low power, embedded platforms
note: any guesses what they could be?
safety of → operator, other people, the vehicle, environment
safety of → operator, other people, the vehicle, environment
a (loose) map of this course