Open-Harbor/osrm-backend
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 3 Wiki Activity

add support for visualising turn penalties in MLD Debug tiles (#4157)

Browse Source 
- template function for tile functionality with edge finder operator
 - refactors unit tests into single function (reduce code duplication)
 - adds unit tests for core-ch
This commit is contained in:
Moritz Kobitzsch 2017-06-15 13:59:44 +02:00 committed by Daniel Patterson
parent 35550d8c0a
commit f80e5db346
6 changed files with 223 additions and 86 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
CHANGELOG.md
View File

@ -18,6 +18,8 @@
- .osrm.nodes file was renamed to .nbg_nodes and .ebg_nodes was added
- Guidance
- #4075 Changed counting of exits on service roundabouts
- Debug Tiles
- added support for visualising turn penalties to the MLD plugin
- Bugfixes
- Fixed a copy/paste issue assigning wrong directions in similar turns (left over right)
- #4074: fixed a bug that would announce entering highway ramps as u-turns

3
include/engine/algorithm.hpp
View File

@ -102,6 +102,9 @@ template <> struct HasShortestPathSearch<mld::Algorithm> final : std::true_type
template <> struct HasMapMatching<mld::Algorithm> final : std::true_type
{
};
template <> struct HasGetTileTurns<mld::Algorithm> final : std::true_type
{
};
}
}
}

9
include/engine/routing_algorithms.hpp
View File

@ -221,15 +221,6 @@ RoutingAlgorithms<routing_algorithms::mld::Algorithm>::ManyToManySearch(
{
throw util::exception("ManyToManySearch is not implemented");
}
template <>
inline std::vector<routing_algorithms::TurnData>
RoutingAlgorithms<routing_algorithms::mld::Algorithm>::GetTileTurns(
const std::vector<datafacade::BaseDataFacade::RTreeLeaf> &,
const std::vector<std::size_t> &) const
{
throw util::exception("GetTileTurns is not implemented");
}
}
}

5
include/engine/routing_algorithms/tile_turns.hpp
View File

@ -33,6 +33,11 @@ getTileTurns(const datafacade::ContiguousInternalMemoryDataFacade<ch::Algorithm>
const std::vector<RTreeLeaf> &edges,
const std::vector<std::size_t> &sorted_edge_indexes);
std::vector<TurnData>
getTileTurns(const datafacade::ContiguousInternalMemoryDataFacade<mld::Algorithm> &facade,
const std::vector<RTreeLeaf> &edges,
const std::vector<std::size_t> &sorted_edge_indexes);
} // namespace routing_algorithms
} // namespace engine
} // namespace osrm

209
src/engine/routing_algorithms/tile_turns.cpp
View File

@ -7,31 +7,33 @@ namespace engine
namespace routing_algorithms
{
std::vector<TurnData>
getTileTurns(const datafacade::ContiguousInternalMemoryDataFacade<ch::Algorithm> &facade,
const std::vector<RTreeLeaf> &edges,
const std::vector<std::size_t> &sorted_edge_indexes)
namespace
{
std::vector<TurnData> all_turn_data;
// Struct to hold info on all the EdgeBasedNodes that are visible in our tile
// When we create these, we insure that (source, target) and packed_geometry_id
// are all pointed in the same direction.
struct EdgeBasedNodeInfo
{
bool is_geometry_forward; // Is the geometry forward or reverse?
unsigned packed_geometry_id;
};
// Struct to hold info on all the EdgeBasedNodes that are visible in our tile
// When we create these, we insure that (source, target) and packed_geometry_id
// are all pointed in the same direction.
struct EdgeBasedNodeInfo
{
bool is_geometry_forward; // Is the geometry forward or reverse?
unsigned packed_geometry_id;
};
struct SegmentData
{
NodeID target_node;
EdgeID edge_based_node_id;
};
template <typename edge_extractor, typename datafacade>
std::vector<TurnData> generateTurns(const datafacade &facade,
const std::vector<RTreeLeaf> &edges,
const std::vector<std::size_t> &sorted_edge_indexes,
edge_extractor const &find_edge)
{
// Lookup table for edge-based-nodes
std::unordered_map<NodeID, EdgeBasedNodeInfo> edge_based_node_info;
struct SegmentData
{
NodeID target_node;
EdgeID edge_based_node_id;
};
std::unordered_map<NodeID, std::vector<SegmentData>> directed_graph;
// Reserve enough space for unique edge-based-nodes on every edge.
// Only a tile with all unique edges will use this much, but
// it saves us a bunch of re-allocations during iteration.
@ -41,8 +43,10 @@ getTileTurns(const datafacade::ContiguousInternalMemoryDataFacade<ch::Algorithm>
return facade.GetGeometryIndex(edge.forward_segment_id.id).id;
};
// Build an adjacency list for all the road segments visible in
// the tile
// To build a tile, we can only rely on the r-tree to quickly find all data visible within the
// tile itself. The Rtree returns a series of segments that may or may not offer turns
// associated with them. To be able to extract turn penalties, we extract a node based graph
// from our edge based representation.
for (const auto &edge_index : sorted_edge_indexes)
{
const auto &edge = edges[edge_index];
@ -79,28 +83,32 @@ getTileTurns(const datafacade::ContiguousInternalMemoryDataFacade<ch::Algorithm>
}
}
// Make sure we traverse the startnodes in a consistent order
// to ensure identical PBF encoding on all platforms.
std::vector<NodeID> sorted_startnodes;
sorted_startnodes.reserve(directed_graph.size());
std::transform(directed_graph.begin(),
directed_graph.end(),
std::back_inserter(sorted_startnodes),
[](auto const &node) { return node.first; });
std::sort(sorted_startnodes.begin(), sorted_startnodes.end());
std::vector<TurnData> all_turn_data;
// Given a turn:
// u---v
// |
// w
// uv is the "approach"
// vw is the "exit"
std::vector<contractor::QueryEdge::EdgeData> unpacked_shortcut;
std::vector<EdgeWeight> approach_weight_vector;
std::vector<EdgeWeight> approach_duration_vector;
// Make sure we traverse the startnodes in a consistent order
// to ensure identical PBF encoding on all platforms.
std::vector<NodeID> sorted_startnodes;
sorted_startnodes.reserve(directed_graph.size());
for (const auto &startnode : directed_graph)
sorted_startnodes.push_back(startnode.first);
std::sort(sorted_startnodes.begin(), sorted_startnodes.end());
// Look at every node in the directed graph we created
for (const auto &startnode : sorted_startnodes)
{
const auto &nodedata = directed_graph[startnode];
BOOST_ASSERT(directed_graph.find(startnode) != directed_graph.end());
const auto &nodedata = directed_graph.find(startnode)->second;
// For all the outgoing edges from the node
for (const auto &approachedge : nodedata)
{
@ -110,7 +118,7 @@ getTileTurns(const datafacade::ContiguousInternalMemoryDataFacade<ch::Algorithm>
continue;
// For each of the outgoing edges from our target coordinate
for (const auto &exit_edge : directed_graph[approachedge.target_node])
for (const auto &exit_edge : directed_graph.find(approachedge.target_node)->second)
{
// If the next edge has the same edge_based_node_id, then it's
// not a turn, so skip it
@ -132,55 +140,32 @@ getTileTurns(const datafacade::ContiguousInternalMemoryDataFacade<ch::Algorithm>
//
// would offer a backward edge at `b` to `a` (due to the oneway from a to b)
// but could also offer a shortcut (b-c-a) from `b` to `a` which is longer.
EdgeID smaller_edge_id =
facade.FindSmallestEdge(approachedge.edge_based_node_id,
exit_edge.edge_based_node_id,
[](const contractor::QueryEdge::EdgeData &data) {
return data.forward && !data.shortcut;
});
EdgeID edge_based_edge_id =
find_edge(approachedge.edge_based_node_id, exit_edge.edge_based_node_id);
// Depending on how the graph is constructed, we might have to look for
// a backwards edge instead. They're equivalent, just one is available for
// a forward routing search, and one is used for the backwards dijkstra
// steps. Their weight should be the same, we can use either one.
// If we didn't find a forward edge, try for a backward one
if (SPECIAL_EDGEID == smaller_edge_id)
if (edge_based_edge_id != SPECIAL_EDGEID)
{
smaller_edge_id =
facade.FindSmallestEdge(exit_edge.edge_based_node_id,
approachedge.edge_based_node_id,
[](const contractor::QueryEdge::EdgeData &data) {
return data.backward && !data.shortcut;
});
}
// If no edge was found, it means that there's no connection between these
// nodes, due to oneways or turn restrictions. Given the edge-based-nodes
// that we're examining here, we *should* only find directly-connected
// edges, not shortcuts
if (smaller_edge_id != SPECIAL_EDGEID)
{
const auto &data = facade.GetEdgeData(smaller_edge_id);
BOOST_ASSERT_MSG(!data.shortcut, "Connecting edge must not be a shortcut");
const auto &data = facade.GetEdgeData(edge_based_edge_id);
// Now, calculate the sum of the weight of all the segments.
if (edge_based_node_info[approachedge.edge_based_node_id].is_geometry_forward)
if (edge_based_node_info.find(approachedge.edge_based_node_id)
->second.is_geometry_forward)
{
approach_weight_vector = facade.GetUncompressedForwardWeights(
edge_based_node_info[approachedge.edge_based_node_id]
.packed_geometry_id);
edge_based_node_info.find(approachedge.edge_based_node_id)
->second.packed_geometry_id);
approach_duration_vector = facade.GetUncompressedForwardDurations(
edge_based_node_info[approachedge.edge_based_node_id]
.packed_geometry_id);
edge_based_node_info.find(approachedge.edge_based_node_id)
->second.packed_geometry_id);
}
else
{
approach_weight_vector = facade.GetUncompressedReverseWeights(
edge_based_node_info[approachedge.edge_based_node_id]
.packed_geometry_id);
edge_based_node_info.find(approachedge.edge_based_node_id)
->second.packed_geometry_id);
approach_duration_vector = facade.GetUncompressedReverseDurations(
edge_based_node_info[approachedge.edge_based_node_id]
.packed_geometry_id);
edge_based_node_info.find(approachedge.edge_based_node_id)
->second.packed_geometry_id);
}
const auto sum_node_weight = std::accumulate(approach_weight_vector.begin(),
approach_weight_vector.end(),
@ -239,6 +224,90 @@ getTileTurns(const datafacade::ContiguousInternalMemoryDataFacade<ch::Algorithm>
return all_turn_data;
}
} // namespace
// CH Version of finding all turn penalties. Here is where the actual work is happening
std::vector<TurnData>
getTileTurns(const datafacade::ContiguousInternalMemoryDataFacade<ch::Algorithm> &facade,
const std::vector<RTreeLeaf> &edges,
const std::vector<std::size_t> &sorted_edge_indexes)
{
// Define how to find the representative edge between two edge based nodes for a CH
struct EdgeFinderCH
{
EdgeFinderCH(const datafacade::ContiguousInternalMemoryDataFacade<ch::Algorithm> &facade)
: facade(facade)
{
}
const datafacade::ContiguousInternalMemoryDataFacade<ch::Algorithm> &facade;
EdgeID operator()(const NodeID approach_node, const NodeID exit_node) const
{
// Find the connection between our source road and the target node
// Since we only want to find direct edges, we cannot check shortcut edges here.
// Otherwise we might find a forward edge even though a shorter backward edge
// exists (due to oneways).
//
// a > - > - > - b
// | |
// |------ c ----|
//
// would offer a backward edge at `b` to `a` (due to the oneway from a to b)
// but could also offer a shortcut (b-c-a) from `b` to `a` which is longer.
EdgeID edge_id = facade.FindSmallestEdge(
approach_node, exit_node, [](const contractor::QueryEdge::EdgeData &data) {
return data.forward && !data.shortcut;
});
// Depending on how the graph is constructed, we might have to look for
// a backwards edge instead. They're equivalent, just one is available for
// a forward routing search, and one is used for the backwards dijkstra
// steps. Their weight should be the same, we can use either one.
// If we didn't find a forward edge, try for a backward one
if (SPECIAL_EDGEID == edge_id)
{
edge_id = facade.FindSmallestEdge(
exit_node, approach_node, [](const contractor::QueryEdge::EdgeData &data) {
return data.backward && !data.shortcut;
});
}
BOOST_ASSERT_MSG(edge_id == SPECIAL_EDGEID || !facade.GetEdgeData(edge_id).shortcut,
"Connecting edge must not be a shortcut");
return edge_id;
}
};
EdgeFinderCH edge_finder(facade);
return generateTurns(facade, edges, sorted_edge_indexes, edge_finder);
}
// MLD version to find all turns
std::vector<TurnData>
getTileTurns(const datafacade::ContiguousInternalMemoryDataFacade<mld::Algorithm> &facade,
const std::vector<RTreeLeaf> &edges,
const std::vector<std::size_t> &sorted_edge_indexes)
{
// Define how to find the representative edge between two edge-based-nodes for a MLD
struct EdgeFinderMLD
{
EdgeFinderMLD(const datafacade::ContiguousInternalMemoryDataFacade<mld::Algorithm> &facade)
: facade(facade)
{
}
const datafacade::ContiguousInternalMemoryDataFacade<mld::Algorithm> &facade;
EdgeID operator()(const NodeID approach_node, const NodeID exit_node) const
{
return facade.FindEdge(approach_node, exit_node);
}
};
EdgeFinderMLD edge_finder(facade);
return generateTurns(facade, edges, sorted_edge_indexes, edge_finder);
}
} // namespace routing_algorithms
} // namespace engine
} // namespace osrm

81
unit_tests/library/tile.cpp
View File

@ -20,12 +20,10 @@
BOOST_AUTO_TEST_SUITE(tile)
BOOST_AUTO_TEST_CASE(test_tile)
template <typename algorithm> void test_tile(algorithm &osrm)
{
using namespace osrm;
auto osrm = getOSRM(OSRM_TEST_DATA_DIR "/ch/monaco.osrm");
// This tile should contain most of monaco
TileParameters params{17059, 11948, 15};
@ -211,11 +209,32 @@ BOOST_AUTO_TEST_CASE(test_tile)
BOOST_CHECK(number_of_turns_found > 700);
}
BOOST_AUTO_TEST_CASE(test_tile_turns)
BOOST_AUTO_TEST_CASE(test_tile_ch)
{
using namespace osrm;
auto osrm = getOSRM(OSRM_TEST_DATA_DIR "/ch/monaco.osrm", osrm::EngineConfig::Algorithm::CH);
test_tile(osrm);
}
BOOST_AUTO_TEST_CASE(test_tile_corech)
{
using namespace osrm;
auto osrm =
getOSRM(OSRM_TEST_DATA_DIR "/corech/monaco.osrm", osrm::EngineConfig::Algorithm::CoreCH);
test_tile(osrm);
}
BOOST_AUTO_TEST_CASE(test_tile_mld)
{
using namespace osrm;
auto osrm = getOSRM(OSRM_TEST_DATA_DIR "/mld/monaco.osrm", osrm::EngineConfig::Algorithm::MLD);
test_tile(osrm);
}
template <typename algorithm> void test_tile_turns(algorithm &osrm)
{
using namespace osrm;
auto osrm = getOSRM(OSRM_TEST_DATA_DIR "/ch/monaco.osrm");
// Small tile where we can test all the values
TileParameters params{272953, 191177, 19};
@ -359,11 +378,34 @@ BOOST_AUTO_TEST_CASE(test_tile_turns)
CHECK_EQUAL_RANGE(actual_turn_bearings, expected_turn_bearings);
}
BOOST_AUTO_TEST_CASE(test_tile_speeds)
BOOST_AUTO_TEST_CASE(test_tile_turns_ch)
{
using namespace osrm;
auto osrm = getOSRM(OSRM_TEST_DATA_DIR "/ch/monaco.osrm", osrm::EngineConfig::Algorithm::CH);
auto osrm = getOSRM(OSRM_TEST_DATA_DIR "/ch/monaco.osrm");
test_tile_turns(osrm);
}
BOOST_AUTO_TEST_CASE(test_tile_turns_corech)
{
using namespace osrm;
auto osrm =
getOSRM(OSRM_TEST_DATA_DIR "/corech/monaco.osrm", osrm::EngineConfig::Algorithm::CoreCH);
test_tile_turns(osrm);
}
BOOST_AUTO_TEST_CASE(test_tile_turns_mld)
{
using namespace osrm;
auto osrm = getOSRM(OSRM_TEST_DATA_DIR "/mld/monaco.osrm", osrm::EngineConfig::Algorithm::MLD);
test_tile_turns(osrm);
}
template <typename algorithm> void test_tile_speeds(algorithm &osrm)
{
using namespace osrm;
// Small tile so we can test all the values
// TileParameters params{272953, 191177, 19};
@ -522,4 +564,29 @@ BOOST_AUTO_TEST_CASE(test_tile_speeds)
BOOST_CHECK(actual_names == expected_names);
}
BOOST_AUTO_TEST_CASE(test_tile_speeds_ch)
{
using namespace osrm;
auto osrm = getOSRM(OSRM_TEST_DATA_DIR "/ch/monaco.osrm", osrm::EngineConfig::Algorithm::CH);
test_tile_speeds(osrm);
}
BOOST_AUTO_TEST_CASE(test_tile_speeds_corech)
{
using namespace osrm;
auto osrm =
getOSRM(OSRM_TEST_DATA_DIR "/corech/monaco.osrm", osrm::EngineConfig::Algorithm::CoreCH);
test_tile_speeds(osrm);
}
BOOST_AUTO_TEST_CASE(test_tile_speeds_mld)
{
using namespace osrm;
auto osrm = getOSRM(OSRM_TEST_DATA_DIR "/mld/monaco.osrm", osrm::EngineConfig::Algorithm::MLD);
test_tile_speeds(osrm);
}
BOOST_AUTO_TEST_SUITE_END()